diff --git a/media/libaaudio/include/aaudio/AAudio.h b/media/libaaudio/include/aaudio/AAudio.h
index 4b08295..212a787 100644
--- a/media/libaaudio/include/aaudio/AAudio.h
+++ b/media/libaaudio/include/aaudio/AAudio.h
@@ -444,6 +444,22 @@
 };
 typedef int32_t aaudio_content_type_t;
 
+enum {
+
+    /**
+     * Constant indicating the audio content associated with these attributes will follow the
+     * default platform behavior with regards to which content will be spatialized or not.
+     */
+    AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO = 1,
+
+    /**
+     * Constant indicating the audio content associated with these attributes should never
+     * be spatialized.
+     */
+    AAUDIO_SPATIALIZATION_BEHAVIOR_NEVER = 2,
+};
+typedef int32_t aaudio_spatialization_behavior_t;
+
 /**
  * Defines the audio source.
  * An audio source defines both a default physical source of audio signal, and a recording
@@ -565,6 +581,145 @@
 };
 typedef int32_t aaudio_session_id_t;
 
+/**
+ * Defines the audio channel mask.
+ * Channel masks are used to describe the samples and their
+ * arrangement in the audio frame. They are also used in the endpoint
+ * (e.g. a USB audio interface, a DAC connected to headphones) to
+ * specify allowable configurations of a particular device.
+ *
+ * Added in API level 32.
+ */
+enum {
+    /**
+     * Invalid channel mask
+     */
+    AAUDIO_CHANNEL_INVALID = -1,
+
+    /**
+     * Output audio channel mask
+     */
+    AAUDIO_CHANNEL_FRONT_LEFT = 1 << 0,
+    AAUDIO_CHANNEL_FRONT_RIGHT = 1 << 1,
+    AAUDIO_CHANNEL_FRONT_CENTER = 1 << 2,
+    AAUDIO_CHANNEL_LOW_FREQUENCY = 1 << 3,
+    AAUDIO_CHANNEL_BACK_LEFT = 1 << 4,
+    AAUDIO_CHANNEL_BACK_RIGHT = 1 << 5,
+    AAUDIO_CHANNEL_FRONT_LEFT_OF_CENTER = 1 << 6,
+    AAUDIO_CHANNEL_FRONT_RIGHT_OF_CENTER = 1 << 7,
+    AAUDIO_CHANNEL_BACK_CENTER = 1 << 8,
+    AAUDIO_CHANNEL_SIDE_LEFT = 1 << 9,
+    AAUDIO_CHANNEL_SIDE_RIGHT = 1 << 10,
+    AAUDIO_CHANNEL_TOP_CENTER = 1 << 11,
+    AAUDIO_CHANNEL_TOP_FRONT_LEFT = 1 << 12,
+    AAUDIO_CHANNEL_TOP_FRONT_CENTER = 1 << 13,
+    AAUDIO_CHANNEL_TOP_FRONT_RIGHT = 1 << 14,
+    AAUDIO_CHANNEL_TOP_BACK_LEFT = 1 << 15,
+    AAUDIO_CHANNEL_TOP_BACK_CENTER = 1 << 16,
+    AAUDIO_CHANNEL_TOP_BACK_RIGHT = 1 << 17,
+    AAUDIO_CHANNEL_TOP_SIDE_LEFT = 1 << 18,
+    AAUDIO_CHANNEL_TOP_SIDE_RIGHT = 1 << 19,
+    AAUDIO_CHANNEL_BOTTOM_FRONT_LEFT = 1 << 20,
+    AAUDIO_CHANNEL_BOTTOM_FRONT_CENTER = 1 << 21,
+    AAUDIO_CHANNEL_BOTTOM_FRONT_RIGHT = 1 << 22,
+    AAUDIO_CHANNEL_LOW_FREQUENCY_2 = 1 << 23,
+    AAUDIO_CHANNEL_FRONT_WIDE_LEFT = 1 << 24,
+    AAUDIO_CHANNEL_FRONT_WIDE_RIGHT = 1 << 25,
+
+    AAUDIO_CHANNEL_MONO = AAUDIO_CHANNEL_FRONT_LEFT,
+    AAUDIO_CHANNEL_STEREO = AAUDIO_CHANNEL_FRONT_LEFT |
+                            AAUDIO_CHANNEL_FRONT_RIGHT,
+    AAUDIO_CHANNEL_2POINT1 = AAUDIO_CHANNEL_FRONT_LEFT |
+                             AAUDIO_CHANNEL_FRONT_RIGHT |
+                             AAUDIO_CHANNEL_LOW_FREQUENCY,
+    AAUDIO_CHANNEL_TRI = AAUDIO_CHANNEL_FRONT_LEFT |
+                         AAUDIO_CHANNEL_FRONT_RIGHT |
+                         AAUDIO_CHANNEL_FRONT_CENTER,
+    AAUDIO_CHANNEL_TRI_BACK = AAUDIO_CHANNEL_FRONT_LEFT |
+                              AAUDIO_CHANNEL_FRONT_RIGHT |
+                              AAUDIO_CHANNEL_BACK_CENTER,
+    AAUDIO_CHANNEL_3POINT1 = AAUDIO_CHANNEL_FRONT_LEFT |
+                             AAUDIO_CHANNEL_FRONT_RIGHT |
+                             AAUDIO_CHANNEL_FRONT_CENTER |
+                             AAUDIO_CHANNEL_LOW_FREQUENCY,
+    AAUDIO_CHANNEL_2POINT0POINT2 = AAUDIO_CHANNEL_FRONT_LEFT |
+                                   AAUDIO_CHANNEL_FRONT_RIGHT |
+                                   AAUDIO_CHANNEL_TOP_SIDE_LEFT |
+                                   AAUDIO_CHANNEL_TOP_SIDE_RIGHT,
+    AAUDIO_CHANNEL_2POINT1POINT2 = AAUDIO_CHANNEL_2POINT0POINT2 |
+                                   AAUDIO_CHANNEL_LOW_FREQUENCY,
+    AAUDIO_CHANNEL_3POINT0POINT2 = AAUDIO_CHANNEL_FRONT_LEFT |
+                                   AAUDIO_CHANNEL_FRONT_RIGHT |
+                                   AAUDIO_CHANNEL_FRONT_CENTER |
+                                   AAUDIO_CHANNEL_TOP_SIDE_LEFT |
+                                   AAUDIO_CHANNEL_TOP_SIDE_RIGHT,
+    AAUDIO_CHANNEL_3POINT1POINT2 = AAUDIO_CHANNEL_3POINT0POINT2 |
+                                   AAUDIO_CHANNEL_LOW_FREQUENCY,
+    AAUDIO_CHANNEL_QUAD = AAUDIO_CHANNEL_FRONT_LEFT |
+                          AAUDIO_CHANNEL_FRONT_RIGHT |
+                          AAUDIO_CHANNEL_BACK_LEFT |
+                          AAUDIO_CHANNEL_BACK_RIGHT,
+    AAUDIO_CHANNEL_QUAD_SIDE = AAUDIO_CHANNEL_FRONT_LEFT |
+                               AAUDIO_CHANNEL_FRONT_RIGHT |
+                               AAUDIO_CHANNEL_SIDE_LEFT |
+                               AAUDIO_CHANNEL_SIDE_RIGHT,
+    AAUDIO_CHANNEL_SURROUND = AAUDIO_CHANNEL_FRONT_LEFT |
+                              AAUDIO_CHANNEL_FRONT_RIGHT |
+                              AAUDIO_CHANNEL_FRONT_CENTER |
+                              AAUDIO_CHANNEL_BACK_CENTER,
+    AAUDIO_CHANNEL_PENTA = AAUDIO_CHANNEL_QUAD |
+                           AAUDIO_CHANNEL_FRONT_CENTER,
+    // aka 5POINT1_BACK
+    AAUDIO_CHANNEL_5POINT1 = AAUDIO_CHANNEL_FRONT_LEFT |
+                             AAUDIO_CHANNEL_FRONT_RIGHT |
+                             AAUDIO_CHANNEL_FRONT_CENTER |
+                             AAUDIO_CHANNEL_LOW_FREQUENCY |
+                             AAUDIO_CHANNEL_BACK_LEFT |
+                             AAUDIO_CHANNEL_BACK_RIGHT,
+    AAUDIO_CHANNEL_5POINT1_SIDE = AAUDIO_CHANNEL_FRONT_LEFT |
+                                  AAUDIO_CHANNEL_FRONT_RIGHT |
+                                  AAUDIO_CHANNEL_FRONT_CENTER |
+                                  AAUDIO_CHANNEL_LOW_FREQUENCY |
+                                  AAUDIO_CHANNEL_SIDE_LEFT |
+                                  AAUDIO_CHANNEL_SIDE_RIGHT,
+    AAUDIO_CHANNEL_6POINT1 = AAUDIO_CHANNEL_FRONT_LEFT |
+                             AAUDIO_CHANNEL_FRONT_RIGHT |
+                             AAUDIO_CHANNEL_FRONT_CENTER |
+                             AAUDIO_CHANNEL_LOW_FREQUENCY |
+                             AAUDIO_CHANNEL_BACK_LEFT |
+                             AAUDIO_CHANNEL_BACK_RIGHT |
+                             AAUDIO_CHANNEL_BACK_CENTER,
+    AAUDIO_CHANNEL_7POINT1 = AAUDIO_CHANNEL_5POINT1 |
+                             AAUDIO_CHANNEL_SIDE_LEFT |
+                             AAUDIO_CHANNEL_SIDE_RIGHT,
+    AAUDIO_CHANNEL_5POINT1POINT2 = AAUDIO_CHANNEL_5POINT1 |
+                                   AAUDIO_CHANNEL_TOP_SIDE_LEFT |
+                                   AAUDIO_CHANNEL_TOP_SIDE_RIGHT,
+    AAUDIO_CHANNEL_5POINT1POINT4 = AAUDIO_CHANNEL_5POINT1 |
+                                   AAUDIO_CHANNEL_TOP_FRONT_LEFT |
+                                   AAUDIO_CHANNEL_TOP_FRONT_RIGHT |
+                                   AAUDIO_CHANNEL_TOP_BACK_LEFT |
+                                   AAUDIO_CHANNEL_TOP_BACK_RIGHT,
+    AAUDIO_CHANNEL_7POINT1POINT2 = AAUDIO_CHANNEL_7POINT1 |
+                                   AAUDIO_CHANNEL_TOP_SIDE_LEFT |
+                                   AAUDIO_CHANNEL_TOP_SIDE_RIGHT,
+    AAUDIO_CHANNEL_7POINT1POINT4 = AAUDIO_CHANNEL_7POINT1 |
+                                   AAUDIO_CHANNEL_TOP_FRONT_LEFT |
+                                   AAUDIO_CHANNEL_TOP_FRONT_RIGHT |
+                                   AAUDIO_CHANNEL_TOP_BACK_LEFT |
+                                   AAUDIO_CHANNEL_TOP_BACK_RIGHT,
+    AAUDIO_CHANNEL_9POINT1POINT4 = AAUDIO_CHANNEL_7POINT1POINT4 |
+                                   AAUDIO_CHANNEL_FRONT_WIDE_LEFT |
+                                   AAUDIO_CHANNEL_FRONT_WIDE_RIGHT,
+    AAUDIO_CHANNEL_9POINT1POINT6 = AAUDIO_CHANNEL_9POINT1POINT4 |
+                                   AAUDIO_CHANNEL_TOP_SIDE_LEFT |
+                                   AAUDIO_CHANNEL_TOP_SIDE_RIGHT,
+
+    AAUDIO_CHANNEL_FRONT_BACK = AAUDIO_CHANNEL_FRONT_CENTER |
+                                AAUDIO_CHANNEL_BACK_CENTER,
+};
+typedef uint32_t aaudio_channel_mask_t;
+
 typedef struct AAudioStreamStruct         AAudioStream;
 typedef struct AAudioStreamBuilderStruct  AAudioStreamBuilder;
 
@@ -699,6 +854,11 @@
  * If an exact value is specified then an opened stream will use that value.
  * If a stream cannot be opened with the specified value then the open will fail.
  *
+ * As the channel count provided here may be different from the corresponding channel count
+ * of channel mask used in {@link AAudioStreamBuilder_setChannelMask}, the last called function
+ * will be respected if both this function and {@link AAudioStreamBuilder_setChannelMask} are
+ * called.
+ *
  * Available since API level 26.
  *
  * @param builder reference provided by AAudio_createStreamBuilder()
@@ -714,6 +874,8 @@
  *
  * @param builder reference provided by AAudio_createStreamBuilder()
  * @param samplesPerFrame Number of samples in a frame.
+ *
+ * @deprecated use {@link AAudioStreamBuilder_setChannelCount}
  */
 AAUDIO_API void AAudioStreamBuilder_setSamplesPerFrame(AAudioStreamBuilder* builder,
                                                        int32_t samplesPerFrame) __INTRODUCED_IN(26);
@@ -836,6 +998,37 @@
         aaudio_content_type_t contentType) __INTRODUCED_IN(28);
 
 /**
+ * Sets the behavior affecting whether spatialization will be used.
+ *
+ * The AAudio system will use this information to select whether the stream will go
+ * through a spatializer effect or not when the effect is supported and enabled.
+ *
+ * Available since API level 32.
+ *
+ * @param builder reference provided by AAudio_createStreamBuilder()
+ * @param spatializationBehavior the desired behavior with regards to spatialization, eg.
+ *     {@link #AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO}
+ */
+AAUDIO_API void AAudioStreamBuilder_setSpatializationBehavior(AAudioStreamBuilder* builder,
+        aaudio_spatialization_behavior_t spatializationBehavior) __INTRODUCED_IN(32);
+
+/**
+ * Specifies whether the audio data of this output stream has already been processed for
+ * spatialization.
+ *
+ * If the stream has been processed for spatialization, setting this to true will prevent
+ * issues such as double-processing on platforms that will spatialize audio data.
+ *
+ * Available since API level 32.
+ *
+ * @param builder reference provided by AAudio_createStreamBuilder()
+ * @param isSpatialized true if the content is already processed for binaural or transaural spatial
+ *     rendering, false otherwise.
+ */
+AAUDIO_API void AAudioStreamBuilder_setIsContentSpatialized(AAudioStreamBuilder* builder,
+        bool isSpatialized) __INTRODUCED_IN(32);
+
+/**
  * Set the input (capture) preset for the stream.
  *
  * The AAudio system will use this information to optimize the
@@ -1136,6 +1329,32 @@
 AAUDIO_API aaudio_result_t  AAudioStreamBuilder_delete(AAudioStreamBuilder* builder)
     __INTRODUCED_IN(26);
 
+/**
+ * Set audio channel mask for the stream.
+ *
+ * The default, if you do not call this function, is {@link #AAUDIO_UNSPECIFIED}.
+ * If both channel mask and count are not set, then stereo will then be chosen when the
+ * stream is opened.
+ * After opening a stream with an unspecified value, the application must query for the
+ * actual value, which may vary by device.
+ *
+ * If an exact value is specified then an opened stream will use that value.
+ * If a stream cannot be opened with the specified value then the open will fail.
+ *
+ * As the corresponding channel count of provided channel mask here may be different
+ * from the channel count used in {@link AAudioStreamBuilder_setChannelCount} or
+ * {@link AAudioStreamBuilder_setSamplesPerFrame}, the last called function will be
+ * respected if this function and {@link AAudioStreamBuilder_setChannelCount} or
+ * {@link AAudioStreamBuilder_setSamplesPerFrame} are called.
+ *
+ * Available since API level 32.
+ *
+ * @param builder reference provided by AAudio_createStreamBuilder()
+ * @param channelMask Audio channel mask desired.
+ */
+AAUDIO_API void AAudioStreamBuilder_setChannelMask(AAudioStreamBuilder* builder,
+        aaudio_channel_mask_t channelMask) __INTRODUCED_IN(32);
+
 // ============================================================
 // Stream Control
 // ============================================================
@@ -1616,6 +1835,31 @@
         __INTRODUCED_IN(28);
 
 /**
+ * Return the spatialization behavior for the stream.
+ *
+ * If none was explicitly set, it will return the default
+ * {@link #AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO} behavior.
+ *
+ * Available since API level 32.
+ *
+ * @param stream reference provided by AAudioStreamBuilder_openStream()
+ * @return spatialization behavior, for example {@link #AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO}
+ */
+AAUDIO_API aaudio_spatialization_behavior_t AAudioStream_getSpatializationBehavior(
+        AAudioStream* stream) __INTRODUCED_IN(32);
+
+/**
+ * Return whether the content of the stream is spatialized.
+ *
+ * Available since API level 32.
+ *
+ * @param stream reference provided by AAudioStreamBuilder_openStream()
+ * @return true if the content is spatialized
+ */
+AAUDIO_API bool AAudioStream_isContentSpatialized(AAudioStream* stream) __INTRODUCED_IN(32);
+
+
+/**
  * Return the input preset for the stream.
  *
  * Available since API level 28.
@@ -1652,6 +1896,18 @@
 AAUDIO_API bool AAudioStream_isPrivacySensitive(AAudioStream* stream)
         __INTRODUCED_IN(30);
 
+/**
+ * Return the channel mask for the stream. This will be the mask set using
+ * {@link #AAudioStreamBuilder_setChannelMask}, or {@link #AAUDIO_UNSPECIFIED} otherwise.
+ *
+ * Available since API level 32.
+ *
+ * @param stream reference provided by AAudioStreamBuilder_openStream()
+ * @return actual channel mask
+ */
+AAUDIO_API aaudio_channel_mask_t AAudioStream_getChannelMask(AAudioStream* stream)
+        __INTRODUCED_IN(32);
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/media/libaaudio/src/binding/AAudioStreamConfiguration.cpp b/media/libaaudio/src/binding/AAudioStreamConfiguration.cpp
index 2d501ef..bec4393 100644
--- a/media/libaaudio/src/binding/AAudioStreamConfiguration.cpp
+++ b/media/libaaudio/src/binding/AAudioStreamConfiguration.cpp
@@ -30,7 +30,7 @@
 using android::media::audio::common::AudioFormat;
 
 AAudioStreamConfiguration::AAudioStreamConfiguration(const StreamParameters& parcelable) {
-    setSamplesPerFrame(parcelable.samplesPerFrame);
+    setChannelMask(parcelable.channelMask);
     setSampleRate(parcelable.sampleRate);
     setDeviceId(parcelable.deviceId);
     static_assert(sizeof(aaudio_sharing_mode_t) == sizeof(parcelable.sharingMode));
@@ -43,6 +43,13 @@
     setUsage(parcelable.usage);
     static_assert(sizeof(aaudio_content_type_t) == sizeof(parcelable.contentType));
     setContentType(parcelable.contentType);
+
+    static_assert(sizeof(aaudio_spatialization_behavior_t) ==
+            sizeof(parcelable.spatializationBehavior));
+    setSpatializationBehavior(parcelable.spatializationBehavior);
+    setIsContentSpatialized(parcelable.isContentSpatialized);
+
+
     static_assert(sizeof(aaudio_input_preset_t) == sizeof(parcelable.inputPreset));
     setInputPreset(parcelable.inputPreset);
     setBufferCapacity(parcelable.bufferCapacity);
@@ -63,7 +70,7 @@
 
 StreamParameters AAudioStreamConfiguration::parcelable() const {
     StreamParameters result;
-    result.samplesPerFrame = getSamplesPerFrame();
+    result.channelMask = getChannelMask();
     result.sampleRate = getSampleRate();
     result.deviceId = getDeviceId();
     static_assert(sizeof(aaudio_sharing_mode_t) == sizeof(result.sharingMode));
diff --git a/media/libaaudio/src/binding/aidl/aaudio/StreamParameters.aidl b/media/libaaudio/src/binding/aidl/aaudio/StreamParameters.aidl
index b7c4f70..a6541e1 100644
--- a/media/libaaudio/src/binding/aidl/aaudio/StreamParameters.aidl
+++ b/media/libaaudio/src/binding/aidl/aaudio/StreamParameters.aidl
@@ -19,7 +19,7 @@
 import android.media.audio.common.AudioFormat;
 
 parcelable StreamParameters {
-    int                                       samplesPerFrame;  //      = AAUDIO_UNSPECIFIED;
+    int                                       channelMask;  //          = AAUDIO_UNSPECIFIED;
     int                                       sampleRate;  //           = AAUDIO_UNSPECIFIED;
     int                                       deviceId;  //             = AAUDIO_UNSPECIFIED;
     int /* aaudio_sharing_mode_t */           sharingMode;  //          = AAUDIO_SHARING_MODE_SHARED;
@@ -27,6 +27,8 @@
     int /* aaudio_direction_t */              direction;  //            = AAUDIO_DIRECTION_OUTPUT;
     int /* aaudio_usage_t */                  usage;  //                = AAUDIO_UNSPECIFIED;
     int /* aaudio_content_type_t */           contentType;  //          = AAUDIO_UNSPECIFIED;
+    int /* aaudio_spatialization_behavior_t */spatializationBehavior; //= AAUDIO_UNSPECIFIED;
+    boolean                                   isContentSpatialized;  // = false;
     int /* aaudio_input_preset_t */           inputPreset;  //          = AAUDIO_UNSPECIFIED;
     int                                       bufferCapacity;  //       = AAUDIO_UNSPECIFIED;
     int /* aaudio_allowed_capture_policy_t */ allowedCapturePolicy;  // = AAUDIO_UNSPECIFIED;
diff --git a/media/libaaudio/src/client/AudioStreamInternal.cpp b/media/libaaudio/src/client/AudioStreamInternal.cpp
index 6d2d464..f933b29 100644
--- a/media/libaaudio/src/client/AudioStreamInternal.cpp
+++ b/media/libaaudio/src/client/AudioStreamInternal.cpp
@@ -123,12 +123,14 @@
 
     request.getConfiguration().setDeviceId(getDeviceId());
     request.getConfiguration().setSampleRate(getSampleRate());
-    request.getConfiguration().setSamplesPerFrame(getSamplesPerFrame());
     request.getConfiguration().setDirection(getDirection());
     request.getConfiguration().setSharingMode(getSharingMode());
+    request.getConfiguration().setChannelMask(getChannelMask());
 
     request.getConfiguration().setUsage(getUsage());
     request.getConfiguration().setContentType(getContentType());
+    request.getConfiguration().setSpatializationBehavior(getSpatializationBehavior());
+    request.getConfiguration().setIsContentSpatialized(isContentSpatialized());
     request.getConfiguration().setInputPreset(getInputPreset());
     request.getConfiguration().setPrivacySensitive(isPrivacySensitive());
 
@@ -138,7 +140,8 @@
 
     mServiceStreamHandle = mServiceInterface.openStream(request, configurationOutput);
     if (mServiceStreamHandle < 0
-            && request.getConfiguration().getSamplesPerFrame() == 1 // mono?
+            && (request.getConfiguration().getSamplesPerFrame() == 1
+                    || request.getConfiguration().getChannelMask() == AAUDIO_CHANNEL_MONO)
             && getDirection() == AAUDIO_DIRECTION_OUTPUT
             && !isInService()) {
         // if that failed then try switching from mono to stereo if OUTPUT.
@@ -146,7 +149,7 @@
         // that writes to a stereo MMAP stream.
         ALOGD("%s() - openStream() returned %d, try switching from MONO to STEREO",
               __func__, mServiceStreamHandle);
-        request.getConfiguration().setSamplesPerFrame(2); // stereo
+        request.getConfiguration().setChannelMask(AAUDIO_CHANNEL_STEREO);
         mServiceStreamHandle = mServiceInterface.openStream(request, configurationOutput);
     }
     if (mServiceStreamHandle < 0) {
@@ -174,9 +177,10 @@
         goto error;
     }
     // Save results of the open.
-    if (getSamplesPerFrame() == AAUDIO_UNSPECIFIED) {
-        setSamplesPerFrame(configurationOutput.getSamplesPerFrame());
+    if (getChannelMask() == AAUDIO_UNSPECIFIED) {
+        setChannelMask(configurationOutput.getChannelMask());
     }
+
     mDeviceChannelCount = configurationOutput.getSamplesPerFrame();
 
     setSampleRate(configurationOutput.getSampleRate());
@@ -186,6 +190,8 @@
 
     setUsage(configurationOutput.getUsage());
     setContentType(configurationOutput.getContentType());
+    setSpatializationBehavior(configurationOutput.getSpatializationBehavior());
+    setIsContentSpatialized(configurationOutput.isContentSpatialized());
     setInputPreset(configurationOutput.getInputPreset());
 
     // Save device format so we can do format conversion and volume scaling together.
diff --git a/media/libaaudio/src/core/AAudioAudio.cpp b/media/libaaudio/src/core/AAudioAudio.cpp
index d103aca..f07e66e 100644
--- a/media/libaaudio/src/core/AAudioAudio.cpp
+++ b/media/libaaudio/src/core/AAudioAudio.cpp
@@ -128,7 +128,8 @@
                                                        int32_t samplesPerFrame)
 {
     AudioStreamBuilder *streamBuilder = convertAAudioBuilderToStreamBuilder(builder);
-    streamBuilder->setSamplesPerFrame(samplesPerFrame);
+    const aaudio_channel_mask_t channelMask = AAudioConvert_channelCountToMask(samplesPerFrame);
+    streamBuilder->setChannelMask(channelMask);
 }
 
 AAUDIO_API void AAudioStreamBuilder_setDirection(AAudioStreamBuilder* builder,
@@ -166,6 +167,18 @@
     streamBuilder->setContentType(contentType);
 }
 
+AAUDIO_API void AAudioStreamBuilder_setSpatializationBehavior(AAudioStreamBuilder* builder,
+        aaudio_spatialization_behavior_t spatializationBehavior) {
+    AudioStreamBuilder *streamBuilder = convertAAudioBuilderToStreamBuilder(builder);
+    streamBuilder->setSpatializationBehavior(spatializationBehavior);
+}
+
+AAUDIO_API void AAudioStreamBuilder_setIsContentSpatialized(AAudioStreamBuilder* builder,
+                                                            bool isSpatialized) {
+    AudioStreamBuilder *streamBuilder = convertAAudioBuilderToStreamBuilder(builder);
+    streamBuilder->setIsContentSpatialized(isSpatialized);
+}
+
 AAUDIO_API void AAudioStreamBuilder_setInputPreset(AAudioStreamBuilder* builder,
                                                    aaudio_input_preset_t inputPreset) {
     AudioStreamBuilder *streamBuilder = convertAAudioBuilderToStreamBuilder(builder);
@@ -223,6 +236,13 @@
     streamBuilder->setFramesPerDataCallback(frames);
 }
 
+AAUDIO_API void AAudioStreamBuilder_setChannelMask(AAudioStreamBuilder* builder,
+                                                   aaudio_channel_mask_t channelMask)
+{
+    AudioStreamBuilder *streamBuilder = convertAAudioBuilderToStreamBuilder(builder);
+    streamBuilder->setChannelMask(channelMask);
+}
+
 AAUDIO_API aaudio_result_t  AAudioStreamBuilder_openStream(AAudioStreamBuilder* builder,
                                                      AAudioStream** streamPtr)
 {
@@ -495,6 +515,19 @@
     return audioStream->getContentType();
 }
 
+AAUDIO_API aaudio_spatialization_behavior_t AAudioStream_getSpatializationBehavior(
+        AAudioStream* stream)
+{
+    AudioStream *audioStream = convertAAudioStreamToAudioStream(stream);
+    return audioStream->getSpatializationBehavior();
+}
+
+AAUDIO_API bool AAudioStream_isContentSpatialized(AAudioStream* stream)
+{
+    AudioStream *audioStream = convertAAudioStreamToAudioStream(stream);
+    return audioStream->isContentSpatialized();
+}
+
 AAUDIO_API aaudio_input_preset_t AAudioStream_getInputPreset(AAudioStream* stream)
 {
     AudioStream *audioStream = convertAAudioStreamToAudioStream(stream);
@@ -562,3 +595,11 @@
     AudioStream *audioStream = convertAAudioStreamToAudioStream(stream);
     return audioStream->isPrivacySensitive();
 }
+
+AAUDIO_API aaudio_channel_mask_t AAudioStream_getChannelMask(AAudioStream* stream)
+{
+    AudioStream *audioStream = convertAAudioStreamToAudioStream(stream);
+    const aaudio_channel_mask_t channelMask = audioStream->getChannelMask();
+    // Do not return channel index masks as they are not public.
+    return AAudio_isChannelIndexMask(channelMask) ? AAUDIO_UNSPECIFIED : channelMask;
+}
diff --git a/media/libaaudio/src/core/AAudioStreamParameters.cpp b/media/libaaudio/src/core/AAudioStreamParameters.cpp
index acfac24..dc242b8 100644
--- a/media/libaaudio/src/core/AAudioStreamParameters.cpp
+++ b/media/libaaudio/src/core/AAudioStreamParameters.cpp
@@ -44,11 +44,14 @@
     mBufferCapacity       = other.mBufferCapacity;
     mUsage                = other.mUsage;
     mContentType          = other.mContentType;
+    mSpatializationBehavior = other.mSpatializationBehavior;
+    mIsContentSpatialized = other.mIsContentSpatialized;
     mInputPreset          = other.mInputPreset;
     mAllowedCapturePolicy = other.mAllowedCapturePolicy;
     mIsPrivacySensitive   = other.mIsPrivacySensitive;
     mOpPackageName        = other.mOpPackageName;
     mAttributionTag       = other.mAttributionTag;
+    mChannelMask          = other.mChannelMask;
 }
 
 static aaudio_result_t isFormatValid(audio_format_t format) {
@@ -160,6 +163,19 @@
             // break;
     }
 
+    switch (mSpatializationBehavior) {
+        case AAUDIO_UNSPECIFIED:
+        case AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO:
+        case AAUDIO_SPATIALIZATION_BEHAVIOR_NEVER:
+            break; // valid
+        default:
+            ALOGD("spatialization behavior not valid = %d", mSpatializationBehavior);
+            return AAUDIO_ERROR_ILLEGAL_ARGUMENT;
+            // break;
+    }
+
+    // no validation required for mIsContentSpatialized
+
     switch (mInputPreset) {
         case AAUDIO_UNSPECIFIED:
         case AAUDIO_INPUT_PRESET_GENERIC:
@@ -187,7 +203,94 @@
             // break;
     }
 
-    return AAUDIO_OK;
+    return validateChannelMask();
+}
+
+bool AAudioStreamParameters::validateChannelMask() const {
+    if (mChannelMask == AAUDIO_UNSPECIFIED) {
+        return AAUDIO_OK;
+    }
+
+    if (mChannelMask & AAUDIO_CHANNEL_BIT_INDEX) {
+        switch (mChannelMask) {
+            case AAUDIO_CHANNEL_INDEX_MASK_1:
+            case AAUDIO_CHANNEL_INDEX_MASK_2:
+            case AAUDIO_CHANNEL_INDEX_MASK_3:
+            case AAUDIO_CHANNEL_INDEX_MASK_4:
+            case AAUDIO_CHANNEL_INDEX_MASK_5:
+            case AAUDIO_CHANNEL_INDEX_MASK_6:
+            case AAUDIO_CHANNEL_INDEX_MASK_7:
+            case AAUDIO_CHANNEL_INDEX_MASK_8:
+            case AAUDIO_CHANNEL_INDEX_MASK_9:
+            case AAUDIO_CHANNEL_INDEX_MASK_10:
+            case AAUDIO_CHANNEL_INDEX_MASK_11:
+            case AAUDIO_CHANNEL_INDEX_MASK_12:
+            case AAUDIO_CHANNEL_INDEX_MASK_13:
+            case AAUDIO_CHANNEL_INDEX_MASK_14:
+            case AAUDIO_CHANNEL_INDEX_MASK_15:
+            case AAUDIO_CHANNEL_INDEX_MASK_16:
+            case AAUDIO_CHANNEL_INDEX_MASK_17:
+            case AAUDIO_CHANNEL_INDEX_MASK_18:
+            case AAUDIO_CHANNEL_INDEX_MASK_19:
+            case AAUDIO_CHANNEL_INDEX_MASK_20:
+            case AAUDIO_CHANNEL_INDEX_MASK_21:
+            case AAUDIO_CHANNEL_INDEX_MASK_22:
+            case AAUDIO_CHANNEL_INDEX_MASK_23:
+            case AAUDIO_CHANNEL_INDEX_MASK_24:
+                return AAUDIO_OK;
+            default:
+                ALOGD("Invalid channel index mask %#x", mChannelMask);
+                return AAUDIO_ERROR_ILLEGAL_ARGUMENT;
+        }
+    }
+
+    if (getDirection() == AAUDIO_DIRECTION_INPUT) {
+        switch (mChannelMask) {
+            case AAUDIO_CHANNEL_MONO:
+            case AAUDIO_CHANNEL_STEREO:
+            case AAUDIO_CHANNEL_FRONT_BACK:
+            case AAUDIO_CHANNEL_2POINT0POINT2:
+            case AAUDIO_CHANNEL_2POINT1POINT2:
+            case AAUDIO_CHANNEL_3POINT0POINT2:
+            case AAUDIO_CHANNEL_3POINT1POINT2:
+            case AAUDIO_CHANNEL_5POINT1:
+                return AAUDIO_OK;
+            default:
+                ALOGD("Invalid channel mask %#x, IN", mChannelMask);
+                return AAUDIO_ERROR_ILLEGAL_ARGUMENT;
+        }
+    } else {
+        switch (mChannelMask) {
+            case AAUDIO_CHANNEL_MONO:
+            case AAUDIO_CHANNEL_STEREO:
+            case AAUDIO_CHANNEL_2POINT1:
+            case AAUDIO_CHANNEL_TRI:
+            case AAUDIO_CHANNEL_TRI_BACK:
+            case AAUDIO_CHANNEL_3POINT1:
+            case AAUDIO_CHANNEL_2POINT0POINT2:
+            case AAUDIO_CHANNEL_2POINT1POINT2:
+            case AAUDIO_CHANNEL_3POINT0POINT2:
+            case AAUDIO_CHANNEL_3POINT1POINT2:
+            case AAUDIO_CHANNEL_QUAD:
+            case AAUDIO_CHANNEL_QUAD_SIDE:
+            case AAUDIO_CHANNEL_SURROUND:
+            case AAUDIO_CHANNEL_PENTA:
+            case AAUDIO_CHANNEL_5POINT1:
+            case AAUDIO_CHANNEL_5POINT1_SIDE:
+            case AAUDIO_CHANNEL_5POINT1POINT2:
+            case AAUDIO_CHANNEL_5POINT1POINT4:
+            case AAUDIO_CHANNEL_6POINT1:
+            case AAUDIO_CHANNEL_7POINT1:
+            case AAUDIO_CHANNEL_7POINT1POINT2:
+            case AAUDIO_CHANNEL_7POINT1POINT4:
+            case AAUDIO_CHANNEL_9POINT1POINT4:
+            case AAUDIO_CHANNEL_9POINT1POINT6:
+                return AAUDIO_OK;
+            default:
+                ALOGD("Invalid channel mask %#x. OUT", mChannelMask);
+                return AAUDIO_ERROR_ILLEGAL_ARGUMENT;
+        }
+    }
 }
 
 void AAudioStreamParameters::dump() const {
@@ -195,12 +298,15 @@
     ALOGD("mSessionId            = %6d", mSessionId);
     ALOGD("mSampleRate           = %6d", mSampleRate);
     ALOGD("mSamplesPerFrame      = %6d", mSamplesPerFrame);
+    ALOGD("mChannelMask          = %#x", mChannelMask);
     ALOGD("mSharingMode          = %6d", (int)mSharingMode);
     ALOGD("mAudioFormat          = %6d", (int)mAudioFormat);
     ALOGD("mDirection            = %6d", mDirection);
     ALOGD("mBufferCapacity       = %6d", mBufferCapacity);
     ALOGD("mUsage                = %6d", mUsage);
     ALOGD("mContentType          = %6d", mContentType);
+    ALOGD("mSpatializationBehavior = %6d", mSpatializationBehavior);
+    ALOGD("mIsContentSpatialized = %s", mIsContentSpatialized ? "true" : "false");
     ALOGD("mInputPreset          = %6d", mInputPreset);
     ALOGD("mAllowedCapturePolicy = %6d", mAllowedCapturePolicy);
     ALOGD("mIsPrivacySensitive   = %s", mIsPrivacySensitive ? "true" : "false");
diff --git a/media/libaaudio/src/core/AAudioStreamParameters.h b/media/libaaudio/src/core/AAudioStreamParameters.h
index 5737052..fed036b 100644
--- a/media/libaaudio/src/core/AAudioStreamParameters.h
+++ b/media/libaaudio/src/core/AAudioStreamParameters.h
@@ -49,13 +49,6 @@
         return mSamplesPerFrame;
     }
 
-    /**
-     * This is also known as channelCount.
-     */
-    void setSamplesPerFrame(int32_t samplesPerFrame) {
-        mSamplesPerFrame = samplesPerFrame;
-    }
-
     audio_format_t getFormat() const {
         return mAudioFormat;
     }
@@ -104,6 +97,22 @@
         mContentType = contentType;
     }
 
+    aaudio_spatialization_behavior_t getSpatializationBehavior() const {
+        return mSpatializationBehavior;
+    }
+
+    void setSpatializationBehavior(aaudio_spatialization_behavior_t spatializationBehavior) {
+        mSpatializationBehavior = spatializationBehavior;
+    }
+
+    bool isContentSpatialized() const {
+        return mIsContentSpatialized;
+    }
+
+    void setIsContentSpatialized(bool isSpatialized) {
+        mIsContentSpatialized = isSpatialized;
+    }
+
     aaudio_input_preset_t getInputPreset() const {
         return mInputPreset;
     }
@@ -153,6 +162,15 @@
         mAttributionTag = attributionTag;
     }
 
+    aaudio_channel_mask_t getChannelMask() const {
+        return mChannelMask;
+    }
+
+    void setChannelMask(aaudio_channel_mask_t channelMask) {
+        mChannelMask = channelMask;
+        mSamplesPerFrame = AAudioConvert_channelMaskToCount(channelMask);
+    }
+
     /**
      * @return bytes per frame of getFormat()
      */
@@ -171,6 +189,8 @@
     void dump() const;
 
 private:
+    bool validateChannelMask() const;
+
     int32_t                         mSamplesPerFrame      = AAUDIO_UNSPECIFIED;
     int32_t                         mSampleRate           = AAUDIO_UNSPECIFIED;
     int32_t                         mDeviceId             = AAUDIO_UNSPECIFIED;
@@ -179,6 +199,9 @@
     aaudio_direction_t              mDirection            = AAUDIO_DIRECTION_OUTPUT;
     aaudio_usage_t                  mUsage                = AAUDIO_UNSPECIFIED;
     aaudio_content_type_t           mContentType          = AAUDIO_UNSPECIFIED;
+    aaudio_spatialization_behavior_t mSpatializationBehavior
+                                                          = AAUDIO_UNSPECIFIED;
+    bool                            mIsContentSpatialized = false;
     aaudio_input_preset_t           mInputPreset          = AAUDIO_UNSPECIFIED;
     int32_t                         mBufferCapacity       = AAUDIO_UNSPECIFIED;
     aaudio_allowed_capture_policy_t mAllowedCapturePolicy = AAUDIO_UNSPECIFIED;
@@ -186,6 +209,7 @@
     bool                            mIsPrivacySensitive   = false;
     std::optional<std::string>      mOpPackageName        = {};
     std::optional<std::string>      mAttributionTag       = {};
+    aaudio_channel_mask_t           mChannelMask          = AAUDIO_UNSPECIFIED;
 };
 
 } /* namespace aaudio */
diff --git a/media/libaaudio/src/core/AudioStream.cpp b/media/libaaudio/src/core/AudioStream.cpp
index 09d9535..73432af 100644
--- a/media/libaaudio/src/core/AudioStream.cpp
+++ b/media/libaaudio/src/core/AudioStream.cpp
@@ -76,6 +76,7 @@
     // Copy parameters from the Builder because the Builder may be deleted after this call.
     // TODO AudioStream should be a subclass of AudioStreamParameters
     mSamplesPerFrame = builder.getSamplesPerFrame();
+    mChannelMask = builder.getChannelMask();
     mSampleRate = builder.getSampleRate();
     mDeviceId = builder.getDeviceId();
     mFormat = builder.getFormat();
@@ -91,6 +92,12 @@
     if (mContentType == AAUDIO_UNSPECIFIED) {
         mContentType = AAUDIO_CONTENT_TYPE_MUSIC;
     }
+    mSpatializationBehavior = builder.getSpatializationBehavior();
+    // for consistency with other properties, note UNSPECIFIED is the same as AUTO
+    if (mSpatializationBehavior == AAUDIO_UNSPECIFIED) {
+        mSpatializationBehavior = AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO;
+    }
+    mIsContentSpatialized = builder.isContentSpatialized();
     mInputPreset = builder.getInputPreset();
     if (mInputPreset == AAUDIO_UNSPECIFIED) {
         mInputPreset = AAUDIO_INPUT_PRESET_VOICE_RECOGNITION;
diff --git a/media/libaaudio/src/core/AudioStream.h b/media/libaaudio/src/core/AudioStream.h
index 9835c8c..7896e63 100644
--- a/media/libaaudio/src/core/AudioStream.h
+++ b/media/libaaudio/src/core/AudioStream.h
@@ -253,6 +253,14 @@
         return mContentType;
     }
 
+    aaudio_spatialization_behavior_t getSpatializationBehavior() const {
+        return mSpatializationBehavior;
+    }
+
+    bool isContentSpatialized() const {
+        return mIsContentSpatialized;
+    }
+
     aaudio_input_preset_t getInputPreset() const {
         return mInputPreset;
     }
@@ -270,7 +278,8 @@
     }
 
     /**
-     * This is only valid after setSamplesPerFrame() and setFormat() have been called.
+     * This is only valid after setChannelMask() and setFormat()
+     * have been called.
      */
     int32_t getBytesPerFrame() const {
         return mSamplesPerFrame * getBytesPerSample();
@@ -284,7 +293,7 @@
     }
 
     /**
-     * This is only valid after setSamplesPerFrame() and setDeviceFormat() have been called.
+     * This is only valid after setChannelMask() and setDeviceFormat() have been called.
      */
     int32_t getBytesPerDeviceFrame() const {
         return getSamplesPerFrame() * audio_bytes_per_sample(getDeviceFormat());
@@ -318,6 +327,15 @@
         return mFramesPerDataCallback;
     }
 
+    aaudio_channel_mask_t getChannelMask() const {
+        return mChannelMask;
+    }
+
+    void setChannelMask(aaudio_channel_mask_t channelMask) {
+        mChannelMask = channelMask;
+        mSamplesPerFrame = AAudioConvert_channelMaskToCount(channelMask);
+    }
+
     /**
      * @return true if data callback has been specified
      */
@@ -495,11 +513,6 @@
     }
 
     // This should not be called after the open() call.
-    void setSamplesPerFrame(int32_t samplesPerFrame) {
-        mSamplesPerFrame = samplesPerFrame;
-    }
-
-    // This should not be called after the open() call.
     void setFramesPerBurst(int32_t framesPerBurst) {
         mFramesPerBurst = framesPerBurst;
     }
@@ -589,6 +602,14 @@
         mContentType = contentType;
     }
 
+    void setSpatializationBehavior(aaudio_spatialization_behavior_t spatializationBehavior) {
+        mSpatializationBehavior = spatializationBehavior;
+    }
+
+    void setIsContentSpatialized(bool isContentSpatialized) {
+        mIsContentSpatialized = isContentSpatialized;
+    }
+
     /**
      * This should not be called after the open() call.
      */
@@ -633,6 +654,7 @@
 
     // These do not change after open().
     int32_t                     mSamplesPerFrame = AAUDIO_UNSPECIFIED;
+    aaudio_channel_mask_t       mChannelMask = AAUDIO_UNSPECIFIED;
     int32_t                     mSampleRate = AAUDIO_UNSPECIFIED;
     int32_t                     mDeviceId = AAUDIO_UNSPECIFIED;
     aaudio_sharing_mode_t       mSharingMode = AAUDIO_SHARING_MODE_SHARED;
@@ -645,6 +667,8 @@
 
     aaudio_usage_t              mUsage           = AAUDIO_UNSPECIFIED;
     aaudio_content_type_t       mContentType     = AAUDIO_UNSPECIFIED;
+    aaudio_spatialization_behavior_t mSpatializationBehavior = AAUDIO_UNSPECIFIED;
+    bool                        mIsContentSpatialized = false;
     aaudio_input_preset_t       mInputPreset     = AAUDIO_UNSPECIFIED;
     aaudio_allowed_capture_policy_t mAllowedCapturePolicy = AAUDIO_ALLOW_CAPTURE_BY_ALL;
     bool                        mIsPrivacySensitive = false;
diff --git a/media/libaaudio/src/core/AudioStreamBuilder.cpp b/media/libaaudio/src/core/AudioStreamBuilder.cpp
index e015592..5e1e007 100644
--- a/media/libaaudio/src/core/AudioStreamBuilder.cpp
+++ b/media/libaaudio/src/core/AudioStreamBuilder.cpp
@@ -268,8 +268,8 @@
 
 void AudioStreamBuilder::logParameters() const {
     // This is very helpful for debugging in the future. Please leave it in.
-    ALOGI("rate   = %6d, channels  = %d, format   = %d, sharing = %s, dir = %s",
-          getSampleRate(), getSamplesPerFrame(), getFormat(),
+    ALOGI("rate   = %6d, channels  = %d, channelMask = %#x, format   = %d, sharing = %s, dir = %s",
+          getSampleRate(), getSamplesPerFrame(), getChannelMask(), getFormat(),
           AAudio_convertSharingModeToShortText(getSharingMode()),
           AAudio_convertDirectionToText(getDirection()));
     ALOGI("device = %6d, sessionId = %d, perfMode = %d, callback: %s with frames = %d",
diff --git a/media/libaaudio/src/legacy/AudioStreamRecord.cpp b/media/libaaudio/src/legacy/AudioStreamRecord.cpp
index dc66742..fe8fb19 100644
--- a/media/libaaudio/src/legacy/AudioStreamRecord.cpp
+++ b/media/libaaudio/src/legacy/AudioStreamRecord.cpp
@@ -65,11 +65,8 @@
     const audio_session_t sessionId = AAudioConvert_aaudioToAndroidSessionId(requestedSessionId);
 
     // TODO Support UNSPECIFIED in AudioRecord. For now, use stereo if unspecified.
-    int32_t samplesPerFrame = (getSamplesPerFrame() == AAUDIO_UNSPECIFIED)
-                              ? 2 : getSamplesPerFrame();
-    audio_channel_mask_t channelMask = samplesPerFrame <= 2 ?
-                               audio_channel_in_mask_from_count(samplesPerFrame) :
-                               audio_channel_mask_for_index_assignment_from_count(samplesPerFrame);
+    audio_channel_mask_t channelMask =
+            AAudio_getChannelMaskForOpen(getChannelMask(), getSamplesPerFrame(), true /*isInput*/);
 
     size_t frameCount = (builder.getBufferCapacity() == AAUDIO_UNSPECIFIED) ? 0
                         : builder.getBufferCapacity();
@@ -115,7 +112,7 @@
     constexpr int32_t kMostLikelySampleRateForFast = 48000;
     if (getFormat() == AUDIO_FORMAT_PCM_FLOAT
             && perfMode == AAUDIO_PERFORMANCE_MODE_LOW_LATENCY
-            && (samplesPerFrame <= 2) // FAST only for mono and stereo
+            && (audio_channel_count_from_in_mask(channelMask) <= 2) // FAST only for mono and stereo
             && (getSampleRate() == kMostLikelySampleRateForFast
                 || getSampleRate() == AAUDIO_UNSPECIFIED)) {
         setDeviceFormat(AUDIO_FORMAT_PCM_16_BIT);
@@ -228,7 +225,9 @@
             .set(AMEDIAMETRICS_PROP_ENCODINGCLIENT, toString(requestedFormat).c_str()).record();
 
     // Get the actual values from the AudioRecord.
-    setSamplesPerFrame(mAudioRecord->channelCount());
+    setChannelMask(AAudioConvert_androidToAAudioChannelMask(
+            mAudioRecord->channelMask(), true /*isInput*/,
+            AAudio_isChannelIndexMask(getChannelMask())));
     setSampleRate(mAudioRecord->getSampleRate());
     setBufferCapacity(getBufferCapacityFromDevice());
     setFramesPerBurst(getFramesPerBurstFromDevice());
diff --git a/media/libaaudio/src/legacy/AudioStreamTrack.cpp b/media/libaaudio/src/legacy/AudioStreamTrack.cpp
index 1d412c0..17736fc 100644
--- a/media/libaaudio/src/legacy/AudioStreamTrack.cpp
+++ b/media/libaaudio/src/legacy/AudioStreamTrack.cpp
@@ -66,13 +66,8 @@
     const aaudio_session_id_t requestedSessionId = builder.getSessionId();
     const audio_session_t sessionId = AAudioConvert_aaudioToAndroidSessionId(requestedSessionId);
 
-    // Try to create an AudioTrack
-    // Use stereo if unspecified.
-    int32_t samplesPerFrame = (getSamplesPerFrame() == AAUDIO_UNSPECIFIED)
-                              ? 2 : getSamplesPerFrame();
-    audio_channel_mask_t channelMask = samplesPerFrame <= 2 ?
-                            audio_channel_out_mask_from_count(samplesPerFrame) :
-                            audio_channel_mask_for_index_assignment_from_count(samplesPerFrame);
+    audio_channel_mask_t channelMask =
+            AAudio_getChannelMaskForOpen(getChannelMask(), getSamplesPerFrame(), false /*isInput*/);
 
     audio_output_flags_t flags;
     aaudio_performance_mode_t perfMode = getPerformanceMode();
@@ -140,7 +135,9 @@
     const audio_usage_t usage =
             AAudioConvert_usageToInternal(builder.getUsage());
     const audio_flags_mask_t attributesFlags =
-        AAudioConvert_allowCapturePolicyToAudioFlagsMask(builder.getAllowedCapturePolicy());
+        AAudioConvert_allowCapturePolicyToAudioFlagsMask(builder.getAllowedCapturePolicy(),
+                                                         builder.getSpatializationBehavior(),
+                                                         builder.isContentSpatialized());
 
     const audio_attributes_t attributes = {
             .content_type = contentType,
@@ -199,7 +196,9 @@
     doSetVolume();
 
     // Get the actual values from the AudioTrack.
-    setSamplesPerFrame(mAudioTrack->channelCount());
+    setChannelMask(AAudioConvert_androidToAAudioChannelMask(
+        mAudioTrack->channelMask(), false /*isInput*/,
+        AAudio_isChannelIndexMask(getChannelMask())));
     setFormat(mAudioTrack->format());
     setDeviceFormat(mAudioTrack->format());
     setSampleRate(mAudioTrack->getSampleRate());
diff --git a/media/libaaudio/src/libaaudio.map.txt b/media/libaaudio/src/libaaudio.map.txt
index 1dd44d1..f45b816 100644
--- a/media/libaaudio/src/libaaudio.map.txt
+++ b/media/libaaudio/src/libaaudio.map.txt
@@ -25,6 +25,9 @@
     AAudioStreamBuilder_setPrivacySensitive;   # introduced=30
     AAudioStreamBuilder_setPackageName;   # introduced=31
     AAudioStreamBuilder_setAttributionTag;   # introduced=31
+    AAudioStreamBuilder_setChannelMask;    # introduced=32
+    AAudioStreamBuilder_setSpatializationBehavior; # introduced=32
+    AAudioStreamBuilder_setIsContentSpatialized;   # introduced=32
     AAudioStreamBuilder_openStream;
     AAudioStreamBuilder_delete;
     AAudioStream_close;
@@ -61,6 +64,9 @@
     AAudioStream_isMMapUsed;
     AAudioStream_isPrivacySensitive;   # introduced=30
     AAudioStream_release;        # introduced=30
+    AAudioStream_getChannelMask;  # introduced=32
+    AAudioStream_getSpatializationBehavior;  # introduced=32
+    AAudioStream_isContentSpatialized;       # introduced=32
   local:
     *;
 };
diff --git a/media/libaaudio/src/utility/AAudioUtilities.cpp b/media/libaaudio/src/utility/AAudioUtilities.cpp
index d795725..e44ccee 100644
--- a/media/libaaudio/src/utility/AAudioUtilities.cpp
+++ b/media/libaaudio/src/utility/AAudioUtilities.cpp
@@ -235,20 +235,46 @@
 }
 
 audio_flags_mask_t AAudioConvert_allowCapturePolicyToAudioFlagsMask(
-        aaudio_allowed_capture_policy_t policy) {
+        aaudio_allowed_capture_policy_t policy,
+        aaudio_spatialization_behavior_t spatializationBehavior,
+        bool isContentSpatialized) {
+    audio_flags_mask_t flagsMask = AUDIO_FLAG_NONE;
     switch (policy) {
         case AAUDIO_UNSPECIFIED:
         case AAUDIO_ALLOW_CAPTURE_BY_ALL:
-            return AUDIO_FLAG_NONE;
+            // flagsMask is not modified
+            break;
         case AAUDIO_ALLOW_CAPTURE_BY_SYSTEM:
-            return AUDIO_FLAG_NO_MEDIA_PROJECTION;
+            flagsMask = static_cast<audio_flags_mask_t>(flagsMask | AUDIO_FLAG_NO_MEDIA_PROJECTION);
+            break;
         case AAUDIO_ALLOW_CAPTURE_BY_NONE:
-            return static_cast<audio_flags_mask_t>(
+            flagsMask = static_cast<audio_flags_mask_t>(flagsMask |
                     AUDIO_FLAG_NO_MEDIA_PROJECTION | AUDIO_FLAG_NO_SYSTEM_CAPTURE);
+            break;
         default:
-            ALOGE("%s() 0x%08X unrecognized", __func__, policy);
-            return AUDIO_FLAG_NONE; //
+            ALOGE("%s() 0x%08X unrecognized capture policy", __func__, policy);
+            // flagsMask is not modified
     }
+
+    switch (spatializationBehavior) {
+        case AAUDIO_UNSPECIFIED:
+        case AAUDIO_SPATIALIZATION_BEHAVIOR_AUTO:
+            // flagsMask is not modified
+            break;
+        case AAUDIO_SPATIALIZATION_BEHAVIOR_NEVER:
+            flagsMask = static_cast<audio_flags_mask_t>(flagsMask | AUDIO_FLAG_NEVER_SPATIALIZE);
+            break;
+        default:
+            ALOGE("%s() 0x%08X unrecognized spatialization behavior",
+                  __func__, spatializationBehavior);
+            // flagsMask is not modified
+    }
+
+    if (isContentSpatialized) {
+        flagsMask = static_cast<audio_flags_mask_t>(flagsMask | AUDIO_FLAG_CONTENT_SPATIALIZED);
+    }
+
+    return flagsMask;
 }
 
 audio_flags_mask_t AAudioConvert_privacySensitiveToAudioFlagsMask(
@@ -256,6 +282,248 @@
     return privacySensitive ? AUDIO_FLAG_CAPTURE_PRIVATE : AUDIO_FLAG_NONE;
 }
 
+audio_channel_mask_t AAudioConvert_aaudioToAndroidChannelLayoutMask(
+        aaudio_channel_mask_t channelMask, bool isInput) {
+    if (isInput) {
+        switch (channelMask) {
+            case AAUDIO_CHANNEL_MONO:
+                return AUDIO_CHANNEL_IN_MONO;
+            case AAUDIO_CHANNEL_STEREO:
+                return AUDIO_CHANNEL_IN_STEREO;
+            case AAUDIO_CHANNEL_FRONT_BACK:
+                return AUDIO_CHANNEL_IN_FRONT_BACK;
+            case AAUDIO_CHANNEL_2POINT0POINT2:
+                return AUDIO_CHANNEL_IN_2POINT0POINT2;
+            case AAUDIO_CHANNEL_2POINT1POINT2:
+                return AUDIO_CHANNEL_IN_2POINT1POINT2;
+            case AAUDIO_CHANNEL_3POINT0POINT2:
+                return AUDIO_CHANNEL_IN_3POINT0POINT2;
+            case AAUDIO_CHANNEL_3POINT1POINT2:
+                return AUDIO_CHANNEL_IN_3POINT1POINT2;
+            case AAUDIO_CHANNEL_5POINT1:
+                return AUDIO_CHANNEL_IN_5POINT1;
+            default:
+                ALOGE("%s() %#x unrecognized", __func__, channelMask);
+                return AUDIO_CHANNEL_INVALID;
+        }
+    } else {
+        switch (channelMask) {
+            case AAUDIO_CHANNEL_MONO:
+                return AUDIO_CHANNEL_OUT_MONO;
+            case AAUDIO_CHANNEL_STEREO:
+                return AUDIO_CHANNEL_OUT_STEREO;
+            case AAUDIO_CHANNEL_2POINT1:
+                return AUDIO_CHANNEL_OUT_2POINT1;
+            case AAUDIO_CHANNEL_TRI:
+                return AUDIO_CHANNEL_OUT_TRI;
+            case AAUDIO_CHANNEL_TRI_BACK:
+                return AUDIO_CHANNEL_OUT_TRI_BACK;
+            case AAUDIO_CHANNEL_3POINT1:
+                return AUDIO_CHANNEL_OUT_3POINT1;
+            case AAUDIO_CHANNEL_2POINT0POINT2:
+                return AUDIO_CHANNEL_OUT_2POINT0POINT2;
+            case AAUDIO_CHANNEL_2POINT1POINT2:
+                return AUDIO_CHANNEL_OUT_2POINT1POINT2;
+            case AAUDIO_CHANNEL_3POINT0POINT2:
+                return AUDIO_CHANNEL_OUT_3POINT0POINT2;
+            case AAUDIO_CHANNEL_3POINT1POINT2:
+                return AUDIO_CHANNEL_OUT_3POINT1POINT2;
+            case AAUDIO_CHANNEL_QUAD:
+                return AUDIO_CHANNEL_OUT_QUAD;
+            case AAUDIO_CHANNEL_QUAD_SIDE:
+                return AUDIO_CHANNEL_OUT_QUAD_SIDE;
+            case AAUDIO_CHANNEL_SURROUND:
+                return AUDIO_CHANNEL_OUT_SURROUND;
+            case AAUDIO_CHANNEL_PENTA:
+                return AUDIO_CHANNEL_OUT_PENTA;
+            case AAUDIO_CHANNEL_5POINT1:
+                return AUDIO_CHANNEL_OUT_5POINT1;
+            case AAUDIO_CHANNEL_5POINT1_SIDE:
+                return AUDIO_CHANNEL_OUT_5POINT1_SIDE;
+            case AAUDIO_CHANNEL_5POINT1POINT2:
+                return AUDIO_CHANNEL_OUT_5POINT1POINT2;
+            case AAUDIO_CHANNEL_5POINT1POINT4:
+                return AUDIO_CHANNEL_OUT_5POINT1POINT4;
+            case AAUDIO_CHANNEL_6POINT1:
+                return AUDIO_CHANNEL_OUT_6POINT1;
+            case AAUDIO_CHANNEL_7POINT1:
+                return AUDIO_CHANNEL_OUT_7POINT1;
+            case AAUDIO_CHANNEL_7POINT1POINT2:
+                return AUDIO_CHANNEL_OUT_7POINT1POINT2;
+            case AAUDIO_CHANNEL_7POINT1POINT4:
+                return AUDIO_CHANNEL_OUT_7POINT1POINT4;
+            // TODO: add 9point1point4 and 9point1point6 when they are added in audio-hal-enums.h
+            // case AAUDIO_CHANNEL_9POINT1POINT4:
+            //     return AUDIO_CHANNEL_OUT_9POINT1POINT4;
+            // case AAUDIO_CHANNEL_9POINT1POINT6:
+            //     return AUDIO_CHANNEL_OUT_9POINT1POINT6;
+            default:
+                ALOGE("%s() %#x unrecognized", __func__, channelMask);
+                return AUDIO_CHANNEL_INVALID;
+        }
+    }
+}
+
+aaudio_channel_mask_t AAudioConvert_androidToAAudioChannelLayoutMask(
+        audio_channel_mask_t channelMask, bool isInput) {
+    if (isInput) {
+        switch (channelMask) {
+            case AUDIO_CHANNEL_IN_MONO:
+                return AAUDIO_CHANNEL_MONO;
+            case AUDIO_CHANNEL_IN_STEREO:
+                return AAUDIO_CHANNEL_STEREO;
+            case AUDIO_CHANNEL_IN_FRONT_BACK:
+                return AAUDIO_CHANNEL_FRONT_BACK;
+            case AUDIO_CHANNEL_IN_2POINT0POINT2:
+                return AAUDIO_CHANNEL_2POINT0POINT2;
+            case AUDIO_CHANNEL_IN_2POINT1POINT2:
+                return AAUDIO_CHANNEL_2POINT1POINT2;
+            case AUDIO_CHANNEL_IN_3POINT0POINT2:
+                return AAUDIO_CHANNEL_3POINT0POINT2;
+            case AUDIO_CHANNEL_IN_3POINT1POINT2:
+                return AAUDIO_CHANNEL_3POINT1POINT2;
+            case AUDIO_CHANNEL_IN_5POINT1:
+                return AAUDIO_CHANNEL_5POINT1;
+            default:
+                ALOGE("%s() %#x unrecognized", __func__, channelMask);
+                return AAUDIO_CHANNEL_INVALID;
+        }
+    } else {
+        switch (channelMask) {
+            case AUDIO_CHANNEL_OUT_MONO:
+                return AAUDIO_CHANNEL_MONO;
+            case AUDIO_CHANNEL_OUT_STEREO:
+                return AAUDIO_CHANNEL_STEREO;
+            case AUDIO_CHANNEL_OUT_2POINT1:
+                return AAUDIO_CHANNEL_2POINT1;
+            case AUDIO_CHANNEL_OUT_TRI:
+                return AAUDIO_CHANNEL_TRI;
+            case AUDIO_CHANNEL_OUT_TRI_BACK:
+                return AAUDIO_CHANNEL_TRI_BACK;
+            case AUDIO_CHANNEL_OUT_3POINT1:
+                return AAUDIO_CHANNEL_3POINT1;
+            case AUDIO_CHANNEL_OUT_2POINT0POINT2:
+                return AAUDIO_CHANNEL_2POINT0POINT2;
+            case AUDIO_CHANNEL_OUT_2POINT1POINT2:
+                return AAUDIO_CHANNEL_2POINT1POINT2;
+            case AUDIO_CHANNEL_OUT_3POINT0POINT2:
+                return AAUDIO_CHANNEL_3POINT0POINT2;
+            case AUDIO_CHANNEL_OUT_3POINT1POINT2:
+                return AAUDIO_CHANNEL_3POINT1POINT2;
+            case AUDIO_CHANNEL_OUT_QUAD:
+                return AAUDIO_CHANNEL_QUAD;
+            case AUDIO_CHANNEL_OUT_QUAD_SIDE:
+                return AAUDIO_CHANNEL_QUAD_SIDE;
+            case AUDIO_CHANNEL_OUT_SURROUND:
+                return AAUDIO_CHANNEL_SURROUND;
+            case AUDIO_CHANNEL_OUT_PENTA:
+                return AAUDIO_CHANNEL_PENTA;
+            case AUDIO_CHANNEL_OUT_5POINT1:
+                return AAUDIO_CHANNEL_5POINT1;
+            case AUDIO_CHANNEL_OUT_5POINT1_SIDE:
+                return AAUDIO_CHANNEL_5POINT1_SIDE;
+            case AUDIO_CHANNEL_OUT_5POINT1POINT2:
+                return AAUDIO_CHANNEL_5POINT1POINT2;
+            case AUDIO_CHANNEL_OUT_5POINT1POINT4:
+                return AAUDIO_CHANNEL_5POINT1POINT4;
+            case AUDIO_CHANNEL_OUT_6POINT1:
+                return AAUDIO_CHANNEL_6POINT1;
+            case AUDIO_CHANNEL_OUT_7POINT1:
+                return AAUDIO_CHANNEL_7POINT1;
+            case AUDIO_CHANNEL_OUT_7POINT1POINT2:
+                return AAUDIO_CHANNEL_7POINT1POINT2;
+            case AUDIO_CHANNEL_OUT_7POINT1POINT4:
+                return AAUDIO_CHANNEL_7POINT1POINT4;
+            // TODO: add 9point1point4 and 9point1point6 when they are added in audio-hal-enums.h
+            // case AUDIO_CHANNEL_OUT_9POINT1POINT4:
+            //     return AAUDIO_CHANNEL_9POINT1POINT4;
+            // case AUDIO_CHANNEL_OUT_9POINT1POINT6:
+            //     return AAUDIO_CHANNEL_9POINT1POINT6;
+            default:
+                ALOGE("%s() %#x unrecognized", __func__, channelMask);
+                return AAUDIO_CHANNEL_INVALID;
+        }
+    }
+}
+
+int32_t AAudioConvert_channelMaskToCount(aaudio_channel_mask_t channelMask) {
+    return __builtin_popcount(channelMask & ~AAUDIO_CHANNEL_BIT_INDEX);
+}
+
+aaudio_channel_mask_t AAudioConvert_channelCountToMask(int32_t channelCount) {
+    if (channelCount < 0 || channelCount > AUDIO_CHANNEL_COUNT_MAX) {
+        return AAUDIO_CHANNEL_INVALID;
+    }
+
+    if (channelCount == 0) {
+        return AAUDIO_UNSPECIFIED;
+    }
+
+    // Return index mask if the channel count is greater than 2.
+    return AAUDIO_CHANNEL_BIT_INDEX | ((1 << channelCount) - 1);
+}
+
+aaudio_channel_mask_t AAudioConvert_androidToAAudioChannelIndexMask(
+        audio_channel_mask_t channelMask) {
+    if (audio_channel_mask_get_representation(channelMask) != AUDIO_CHANNEL_REPRESENTATION_INDEX) {
+        ALOGE("%s() %#x not an index mask", __func__, channelMask);
+        return AAUDIO_CHANNEL_INVALID;
+    }
+    return (channelMask & ~AUDIO_CHANNEL_INDEX_HDR) | AAUDIO_CHANNEL_BIT_INDEX;
+}
+
+audio_channel_mask_t AAudioConvert_aaudioToAndroidChannelIndexMask(
+        aaudio_channel_mask_t channelMask) {
+    if (!AAudio_isChannelIndexMask(channelMask)) {
+        ALOGE("%s() %#x not an index mask", __func__, channelMask);
+        return AUDIO_CHANNEL_INVALID;
+    }
+    return audio_channel_mask_for_index_assignment_from_count(
+            AAudioConvert_channelMaskToCount(channelMask));
+}
+
+aaudio_channel_mask_t AAudioConvert_androidToAAudioChannelMask(
+        audio_channel_mask_t channelMask, bool isInput, bool indexMaskRequired) {
+    if (audio_channel_mask_get_representation(channelMask) == AUDIO_CHANNEL_REPRESENTATION_INDEX) {
+        return AAudioConvert_androidToAAudioChannelIndexMask(channelMask);
+    }
+    if (indexMaskRequired) {
+        // Require index mask, `channelMask` here is a position mask.
+        const int channelCount = isInput ? audio_channel_count_from_in_mask(channelMask)
+                                         : audio_channel_count_from_out_mask(channelMask);
+        return AAudioConvert_channelCountToMask(channelCount);
+    }
+    return AAudioConvert_androidToAAudioChannelLayoutMask(channelMask, isInput);
+}
+
+audio_channel_mask_t AAudioConvert_aaudioToAndroidChannelMask(
+        aaudio_channel_mask_t channelMask, bool isInput) {
+    return AAudio_isChannelIndexMask(channelMask)
+            ? AAudioConvert_aaudioToAndroidChannelIndexMask(channelMask)
+            : AAudioConvert_aaudioToAndroidChannelLayoutMask(channelMask, isInput);
+}
+
+bool AAudio_isChannelIndexMask(aaudio_channel_mask_t channelMask) {
+    return (channelMask & AAUDIO_CHANNEL_BIT_INDEX) == AAUDIO_CHANNEL_BIT_INDEX;
+}
+
+audio_channel_mask_t AAudio_getChannelMaskForOpen(
+        aaudio_channel_mask_t channelMask, int32_t samplesPerFrame, bool isInput) {
+    if (channelMask != AAUDIO_UNSPECIFIED) {
+        if (AAudio_isChannelIndexMask(channelMask) && samplesPerFrame <= 2) {
+            // When it is index mask and the count is less than 3, use position mask
+            // instead of index mask for opening a stream. This may need to be revisited
+            // when making channel index mask public.
+            return isInput ? audio_channel_in_mask_from_count(samplesPerFrame)
+                           : audio_channel_out_mask_from_count(samplesPerFrame);
+        }
+        return AAudioConvert_aaudioToAndroidChannelMask(channelMask, isInput);
+    }
+
+    // Return stereo when unspecified.
+    return isInput ? AUDIO_CHANNEL_IN_STEREO : AUDIO_CHANNEL_OUT_STEREO;
+}
+
 int32_t AAudioConvert_framesToBytes(int32_t numFrames,
                                     int32_t bytesPerFrame,
                                     int32_t *sizeInBytes) {
diff --git a/media/libaaudio/src/utility/AAudioUtilities.h b/media/libaaudio/src/utility/AAudioUtilities.h
index 82eb77d..f24df46 100644
--- a/media/libaaudio/src/utility/AAudioUtilities.h
+++ b/media/libaaudio/src/utility/AAudioUtilities.h
@@ -91,11 +91,40 @@
  * @return internal audio flags mask
  */
 audio_flags_mask_t AAudioConvert_allowCapturePolicyToAudioFlagsMask(
-        aaudio_allowed_capture_policy_t policy);
+        aaudio_allowed_capture_policy_t policy,
+        aaudio_spatialization_behavior_t spatializationBehavior,
+        bool isContentSpatialized);
 
 audio_flags_mask_t AAudioConvert_privacySensitiveToAudioFlagsMask(
         bool privacySensitive);
 
+audio_channel_mask_t AAudioConvert_aaudioToAndroidChannelLayoutMask(
+        aaudio_channel_mask_t channelMask, bool isInput);
+
+aaudio_channel_mask_t AAudioConvert_androidToAAudioChannelLayoutMask(
+        audio_channel_mask_t channelMask, bool isInput);
+
+aaudio_channel_mask_t AAudioConvert_androidToAAudioChannelIndexMask(
+        audio_channel_mask_t channelMask);
+
+audio_channel_mask_t AAudioConvert_aaudioToAndroidChannelIndexMask(
+        aaudio_channel_mask_t channelMask);
+
+aaudio_channel_mask_t AAudioConvert_androidToAAudioChannelMask(
+        audio_channel_mask_t channelMask, bool isInput, bool indexMaskRequired);
+
+audio_channel_mask_t AAudioConvert_aaudioToAndroidChannelMask(
+        aaudio_channel_mask_t channelMask, bool isInput);
+
+bool AAudio_isChannelIndexMask(aaudio_channel_mask_t channelMask);
+
+int32_t AAudioConvert_channelMaskToCount(aaudio_channel_mask_t channelMask);
+
+aaudio_channel_mask_t AAudioConvert_channelCountToMask(int32_t channelCount);
+
+audio_channel_mask_t AAudio_getChannelMaskForOpen(
+        aaudio_channel_mask_t channelMask, int32_t samplesPerFrame, bool isInput);
+
 // Note that this code may be replaced by Settings or by some other system configuration tool.
 
 /**
@@ -318,4 +347,36 @@
     std::atomic<int> mRequested{0};
     std::atomic<int> mAcknowledged{0};
 };
+
+enum {
+    /**
+     * Audio channel index mask, only used internally.
+     */
+    AAUDIO_CHANNEL_BIT_INDEX = 0x80000000,
+    AAUDIO_CHANNEL_INDEX_MASK_1 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 1) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_2 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 2) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_3 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 3) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_4 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 4) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_5 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 5) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_6 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 6) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_7 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 7) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_8 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 8) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_9 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 9) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_10 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 10) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_11 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 11) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_12 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 12) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_13 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 13) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_14 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 14) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_15 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 15) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_16 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 16) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_17 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 17) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_18 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 18) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_19 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 19) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_20 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 20) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_21 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 21) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_22 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 22) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_23 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 23) - 1,
+    AAUDIO_CHANNEL_INDEX_MASK_24 = AAUDIO_CHANNEL_BIT_INDEX | (1 << 24) - 1,
+};
+
 #endif //UTILITY_AAUDIO_UTILITIES_H
diff --git a/media/libaudioclient/AidlConversion.cpp b/media/libaudioclient/AidlConversion.cpp
index 321e7f9..acab774 100644
--- a/media/libaudioclient/AidlConversion.cpp
+++ b/media/libaudioclient/AidlConversion.cpp
@@ -1361,6 +1361,10 @@
             return AUDIO_FLAG_NO_SYSTEM_CAPTURE;
         case media::AudioFlag::CAPTURE_PRIVATE:
             return AUDIO_FLAG_CAPTURE_PRIVATE;
+        case media::AudioFlag::CONTENT_SPATIALIZED:
+            return AUDIO_FLAG_CONTENT_SPATIALIZED;
+        case media::AudioFlag::NEVER_SPATIALIZE:
+            return AUDIO_FLAG_NEVER_SPATIALIZE;
     }
     return unexpected(BAD_VALUE);
 }
@@ -1398,6 +1402,10 @@
             return media::AudioFlag::NO_SYSTEM_CAPTURE;
         case AUDIO_FLAG_CAPTURE_PRIVATE:
             return media::AudioFlag::CAPTURE_PRIVATE;
+        case AUDIO_FLAG_CONTENT_SPATIALIZED:
+            return media::AudioFlag::CONTENT_SPATIALIZED;
+        case AUDIO_FLAG_NEVER_SPATIALIZE:
+            return media::AudioFlag::NEVER_SPATIALIZE;
     }
     return unexpected(BAD_VALUE);
 }
diff --git a/media/libaudioclient/Android.bp b/media/libaudioclient/Android.bp
index 9c307ff..3c3d340 100644
--- a/media/libaudioclient/Android.bp
+++ b/media/libaudioclient/Android.bp
@@ -25,11 +25,13 @@
     static_libs: [
         "audioflinger-aidl-cpp",
         "audiopolicy-aidl-cpp",
+        "spatializer-aidl-cpp",
         "av-types-aidl-cpp",
     ],
     export_static_lib_headers: [
         "audioflinger-aidl-cpp",
         "audiopolicy-aidl-cpp",
+        "spatializer-aidl-cpp",
         "av-types-aidl-cpp",
     ],
     target: {
@@ -112,6 +114,7 @@
         "audioclient-types-aidl-cpp",
         "audioflinger-aidl-cpp",
         "audiopolicy-aidl-cpp",
+        "spatializer-aidl-cpp",
         "audiopolicy-types-aidl-cpp",
         "av-types-aidl-cpp",
         "capture_state_listener-aidl-cpp",
@@ -137,6 +140,7 @@
     export_shared_lib_headers: [
         "audioflinger-aidl-cpp",
         "audiopolicy-aidl-cpp",
+        "spatializer-aidl-cpp",
         "framework-permission-aidl-cpp",
         "libbinder",
     ],
@@ -389,6 +393,9 @@
         "aidl/android/media/AudioVolumeGroup.aidl",
         "aidl/android/media/DeviceRole.aidl",
         "aidl/android/media/SoundTriggerSession.aidl",
+        "aidl/android/media/SpatializationLevel.aidl",
+        "aidl/android/media/SpatializationMode.aidl",
+        "aidl/android/media/SpatializerHeadTrackingMode.aidl",
     ],
     imports: [
         "audio_common-aidl",
@@ -459,6 +466,7 @@
     srcs: [
         "aidl/android/media/GetInputForAttrResponse.aidl",
         "aidl/android/media/GetOutputForAttrResponse.aidl",
+        "aidl/android/media/GetSpatializerResponse.aidl",
         "aidl/android/media/Int.aidl",
         "aidl/android/media/RecordClientInfo.aidl",
         "aidl/android/media/IAudioPolicyService.aidl",
@@ -470,6 +478,34 @@
         "audiopolicy-types-aidl",
         "capture_state_listener-aidl",
         "framework-permission-aidl",
+        "spatializer-aidl",
+    ],
+
+    double_loadable: true,
+    backend: {
+        cpp: {
+            min_sdk_version: "29",
+            apex_available: [
+                "//apex_available:platform",
+                "com.android.media",
+            ],
+        },
+    },
+}
+
+aidl_interface {
+    name: "spatializer-aidl",
+    unstable: true,
+    local_include_dir: "aidl",
+    host_supported: true,
+    vendor_available: true,
+    srcs: [
+        "aidl/android/media/INativeSpatializerCallback.aidl",
+        "aidl/android/media/ISpatializer.aidl",
+        "aidl/android/media/ISpatializerHeadTrackingCallback.aidl",
+    ],
+    imports: [
+        "audiopolicy-types-aidl",
     ],
 
     double_loadable: true,
diff --git a/media/libaudioclient/AudioEffect.cpp b/media/libaudioclient/AudioEffect.cpp
index 6ad5483..9091599 100644
--- a/media/libaudioclient/AudioEffect.cpp
+++ b/media/libaudioclient/AudioEffect.cpp
@@ -70,7 +70,8 @@
                 audio_session_t sessionId,
                 audio_io_handle_t io,
                 const AudioDeviceTypeAddr& device,
-                bool probe)
+                bool probe,
+                bool notifyFramesProcessed)
 {
     sp<media::IEffect> iEffect;
     sp<IMemory> cblk;
@@ -124,6 +125,7 @@
     request.device = VALUE_OR_RETURN_STATUS(legacy2aidl_AudioDeviceTypeAddress(device));
     request.attributionSource = mClientAttributionSource;
     request.probe = probe;
+    request.notifyFramesProcessed = notifyFramesProcessed;
 
     media::CreateEffectResponse response;
 
@@ -194,7 +196,8 @@
                 audio_session_t sessionId,
                 audio_io_handle_t io,
                 const AudioDeviceTypeAddr& device,
-                bool probe)
+                bool probe,
+                bool notifyFramesProcessed)
 {
     effect_uuid_t type;
     effect_uuid_t *pType = nullptr;
@@ -211,7 +214,8 @@
         pUuid = &uuid;
     }
 
-    return set(pType, pUuid, priority, cbf, user, sessionId, io, device, probe);
+    return set(pType, pUuid, priority, cbf, user, sessionId, io,
+               device, probe, notifyFramesProcessed);
 }
 
 
@@ -522,6 +526,13 @@
     }
 }
 
+void AudioEffect::framesProcessed(int32_t frames)
+{
+    if (mCbf != NULL) {
+        mCbf(EVENT_FRAMES_PROCESSED, mUserData, &frames);
+    }
+}
+
 // -------------------------------------------------------------------------
 
 status_t AudioEffect::queryNumberEffects(uint32_t *numEffects)
diff --git a/media/libaudioclient/AudioSystem.cpp b/media/libaudioclient/AudioSystem.cpp
index f1eeaa3..c7967e5 100644
--- a/media/libaudioclient/AudioSystem.cpp
+++ b/media/libaudioclient/AudioSystem.cpp
@@ -480,6 +480,12 @@
     return af->systemReady();
 }
 
+status_t AudioSystem::audioPolicyReady() {
+    const sp<IAudioFlinger>& af = AudioSystem::get_audio_flinger();
+    if (af == 0) return NO_INIT;
+    return af->audioPolicyReady();
+}
+
 status_t AudioSystem::getFrameCountHAL(audio_io_handle_t ioHandle,
                                        size_t* frameCount) {
     const sp<IAudioFlinger>& af = AudioSystem::get_audio_flinger();
@@ -2234,6 +2240,47 @@
     return OK;
 }
 
+status_t AudioSystem::getSpatializer(const sp<media::INativeSpatializerCallback>& callback,
+                                          sp<media::ISpatializer>* spatializer) {
+    const sp<IAudioPolicyService>& aps = AudioSystem::get_audio_policy_service();
+    if (spatializer == nullptr) {
+        return BAD_VALUE;
+    }
+    if (aps == 0) {
+        return PERMISSION_DENIED;
+    }
+    media::GetSpatializerResponse response;
+    RETURN_STATUS_IF_ERROR(statusTFromBinderStatus(
+            aps->getSpatializer(callback, &response)));
+
+    *spatializer = response.spatializer;
+    return OK;
+}
+
+status_t AudioSystem::canBeSpatialized(const audio_attributes_t *attr,
+                                    const audio_config_t *config,
+                                    const AudioDeviceTypeAddrVector &devices,
+                                    bool *canBeSpatialized) {
+    const sp<IAudioPolicyService>& aps = AudioSystem::get_audio_policy_service();
+    if (aps == 0) {
+        return PERMISSION_DENIED;
+    }
+    audio_attributes_t attributes = attr != nullptr ? *attr : AUDIO_ATTRIBUTES_INITIALIZER;
+    audio_config_t configuration = config != nullptr ? *config : AUDIO_CONFIG_INITIALIZER;
+
+    std::optional<media::AudioAttributesInternal> attrAidl = VALUE_OR_RETURN_STATUS(
+            legacy2aidl_audio_attributes_t_AudioAttributesInternal(attributes));
+    std::optional<media::AudioConfig> configAidl = VALUE_OR_RETURN_STATUS(
+            legacy2aidl_audio_config_t_AudioConfig(configuration));
+    std::vector<media::AudioDevice> devicesAidl = VALUE_OR_RETURN_STATUS(
+            convertContainer<std::vector<media::AudioDevice>>(devices,
+                                                   legacy2aidl_AudioDeviceTypeAddress));
+    RETURN_STATUS_IF_ERROR(statusTFromBinderStatus(
+            aps->canBeSpatialized(attrAidl, configAidl, devicesAidl, canBeSpatialized)));
+    return OK;
+}
+
+
 class CaptureStateListenerImpl : public media::BnCaptureStateListener,
                                  public IBinder::DeathRecipient {
 public:
diff --git a/media/libaudioclient/IAudioFlinger.cpp b/media/libaudioclient/IAudioFlinger.cpp
index cae81f0..2af1c50 100644
--- a/media/libaudioclient/IAudioFlinger.cpp
+++ b/media/libaudioclient/IAudioFlinger.cpp
@@ -715,6 +715,10 @@
     return statusTFromBinderStatus(mDelegate->systemReady());
 }
 
+status_t AudioFlingerClientAdapter::audioPolicyReady() {
+    return statusTFromBinderStatus(mDelegate->audioPolicyReady());
+}
+
 size_t AudioFlingerClientAdapter::frameCountHAL(audio_io_handle_t ioHandle) const {
     auto result = [&]() -> ConversionResult<size_t> {
         int32_t ioHandleAidl = VALUE_OR_RETURN(legacy2aidl_audio_io_handle_t_int32_t(ioHandle));
@@ -1189,6 +1193,11 @@
     return Status::fromStatusT(mDelegate->systemReady());
 }
 
+Status AudioFlingerServerAdapter::audioPolicyReady() {
+    mDelegate->audioPolicyReady();
+    return Status::ok();
+}
+
 Status AudioFlingerServerAdapter::frameCountHAL(int32_t ioHandle, int64_t* _aidl_return) {
     audio_io_handle_t ioHandleLegacy = VALUE_OR_RETURN_BINDER(
             aidl2legacy_int32_t_audio_io_handle_t(ioHandle));
diff --git a/media/libaudioclient/aidl/android/media/AudioFlag.aidl b/media/libaudioclient/aidl/android/media/AudioFlag.aidl
index 58b493b..91361fb 100644
--- a/media/libaudioclient/aidl/android/media/AudioFlag.aidl
+++ b/media/libaudioclient/aidl/android/media/AudioFlag.aidl
@@ -34,4 +34,6 @@
     MUTE_HAPTIC = 11,
     NO_SYSTEM_CAPTURE = 12,
     CAPTURE_PRIVATE = 13,
+    CONTENT_SPATIALIZED = 14,
+    NEVER_SPATIALIZE = 15,
 }
diff --git a/media/libaudioclient/aidl/android/media/AudioVibratorInfo.aidl b/media/libaudioclient/aidl/android/media/AudioVibratorInfo.aidl
index f88fc3c..8538d8a 100644
--- a/media/libaudioclient/aidl/android/media/AudioVibratorInfo.aidl
+++ b/media/libaudioclient/aidl/android/media/AudioVibratorInfo.aidl
@@ -24,4 +24,5 @@
     int id;
     float resonantFrequency;
     float qFactor;
+    float maxAmplitude;
 }
diff --git a/media/libaudioclient/aidl/android/media/CreateEffectRequest.aidl b/media/libaudioclient/aidl/android/media/CreateEffectRequest.aidl
index 2d274f4..35a56eb 100644
--- a/media/libaudioclient/aidl/android/media/CreateEffectRequest.aidl
+++ b/media/libaudioclient/aidl/android/media/CreateEffectRequest.aidl
@@ -37,4 +37,6 @@
     AudioDevice device;
     AttributionSourceState attributionSource;
     boolean probe;
+    /** true if a callback must be sent each time audio frames are processed */
+    boolean notifyFramesProcessed;
 }
diff --git a/media/libaudioclient/aidl/android/media/GetSpatializerResponse.aidl b/media/libaudioclient/aidl/android/media/GetSpatializerResponse.aidl
new file mode 100644
index 0000000..25115ac
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/GetSpatializerResponse.aidl
@@ -0,0 +1,28 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+import android.media.ISpatializer;
+
+/**
+ * Used as a return value for IAudioPolicyService.getSpatializer() method
+ * {@hide}
+ */
+ parcelable GetSpatializerResponse {
+    /* The ISpatializer interface if successful, null if not */
+    @nullable ISpatializer spatializer;
+}
diff --git a/media/libaudioclient/aidl/android/media/IAudioFlingerService.aidl b/media/libaudioclient/aidl/android/media/IAudioFlingerService.aidl
index d2cae6d..7ffcc33 100644
--- a/media/libaudioclient/aidl/android/media/IAudioFlingerService.aidl
+++ b/media/libaudioclient/aidl/android/media/IAudioFlingerService.aidl
@@ -197,6 +197,9 @@
     /* Indicate JAVA services are ready (scheduling, power management ...) */
     oneway void systemReady();
 
+    /* Indicate audio policy service is ready */
+    oneway void audioPolicyReady();
+
     // Returns the number of frames per audio HAL buffer.
     long frameCountHAL(int /* audio_io_handle_t */ ioHandle);
 
diff --git a/media/libaudioclient/aidl/android/media/IAudioPolicyService.aidl b/media/libaudioclient/aidl/android/media/IAudioPolicyService.aidl
index 4c3955a..5f0a1de 100644
--- a/media/libaudioclient/aidl/android/media/IAudioPolicyService.aidl
+++ b/media/libaudioclient/aidl/android/media/IAudioPolicyService.aidl
@@ -47,8 +47,10 @@
 import android.media.EffectDescriptor;
 import android.media.GetInputForAttrResponse;
 import android.media.GetOutputForAttrResponse;
+import android.media.GetSpatializerResponse;
 import android.media.IAudioPolicyServiceClient;
 import android.media.ICaptureStateListener;
+import android.media.INativeSpatializerCallback;
 import android.media.Int;
 import android.media.SoundTriggerSession;
 
@@ -348,4 +350,29 @@
                                                     DeviceRole role);
 
     boolean registerSoundTriggerCaptureStateListener(ICaptureStateListener listener);
+
+    /** If a spatializer stage effect is present on the platform, this will return an
+     * ISpatializer interface (see GetSpatializerResponse,aidl) to control this
+     * feature.
+     * If no spatializer stage is present, a null interface is returned.
+     * The INativeSpatializerCallback passed must not be null.
+     * Only one ISpatializer interface can exist at a given time. The native audio policy
+     * service will reject the request if an interface was already acquired and previous owner
+     * did not die or call ISpatializer.release().
+     */
+    GetSpatializerResponse getSpatializer(INativeSpatializerCallback callback);
+
+    /** Queries if some kind of spatialization will be performed if the audio playback context
+     * described by the provided arguments is present.
+     * The context is made of:
+     * - The audio attributes describing the playback use case.
+     * - The audio configuration describing the audio format, channels, sampling rate...
+     * - The devices describing the sink audio device selected for playback.
+     * All arguments are optional and only the specified arguments are used to match against
+     * supported criteria. For instance, supplying no argument will tell if spatialization is
+     * supported or not in general.
+     */
+    boolean canBeSpatialized(in @nullable AudioAttributesInternal attr,
+                             in @nullable AudioConfig config,
+                             in AudioDevice[] devices);
 }
diff --git a/media/libaudioclient/aidl/android/media/IEffectClient.aidl b/media/libaudioclient/aidl/android/media/IEffectClient.aidl
index 3b6bcf1..37b442d 100644
--- a/media/libaudioclient/aidl/android/media/IEffectClient.aidl
+++ b/media/libaudioclient/aidl/android/media/IEffectClient.aidl
@@ -43,4 +43,10 @@
      * TODO(ytai): replace opaque byte arrays with strongly typed parameters.
      */
     oneway void commandExecuted(int cmdCode, in byte[] cmdData, in byte[] replyData);
+
+    /**
+     * Called whenever audio frames have been processed by the effect engine.
+     * @param frames number of frames processed.
+     */
+    oneway void framesProcessed(int frames);
 }
diff --git a/media/libaudioclient/aidl/android/media/INativeSpatializerCallback.aidl b/media/libaudioclient/aidl/android/media/INativeSpatializerCallback.aidl
new file mode 100644
index 0000000..0e9634c
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/INativeSpatializerCallback.aidl
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+import android.media.SpatializationLevel;
+import android.media.SpatializerHeadTrackingMode;
+
+/**
+ * The INativeSpatializerCallback interface is a callback associated to the
+ * ISpatializer interface. The callback is used by the spatializer
+ * implementation in native audio server to communicate state changes to the
+ * client controlling the spatializer with the ISpatializer interface.
+ * {@hide}
+ */
+oneway interface INativeSpatializerCallback {
+    /** Called when the spatialization level applied by the spatializer changes
+     * (e.g. when the spatializer is enabled or disabled)
+     */
+    void onLevelChanged(SpatializationLevel level);
+}
diff --git a/media/libaudioclient/aidl/android/media/ISpatializer.aidl b/media/libaudioclient/aidl/android/media/ISpatializer.aidl
new file mode 100644
index 0000000..b871238
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/ISpatializer.aidl
@@ -0,0 +1,138 @@
+/*
+ * Copyright 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+import android.media.ISpatializerHeadTrackingCallback;
+import android.media.SpatializationLevel;
+import android.media.SpatializationMode;
+import android.media.SpatializerHeadTrackingMode;
+
+
+/**
+ * The ISpatializer interface is used to control the native audio service implementation
+ * of the spatializer stage with headtracking when present on a platform.
+ * It is intended for exclusive use by the java AudioService running in system_server.
+ * It provides APIs to discover the feature availability and options as well as control and report
+ * the active state and modes of the spatializer and head tracking effect.
+ * {@hide}
+ */
+interface ISpatializer {
+    /** Releases a ISpatializer interface previously acquired. */
+    void release();
+
+    /** Reports the list of supported spatialization levels (see SpatializationLevel.aidl).
+     * The list should never be empty if an ISpatializer interface was successfully
+     * retrieved with IAudioPolicyService.getSpatializer().
+     */
+    SpatializationLevel[] getSupportedLevels();
+
+    /** Selects the desired spatialization level (see SpatializationLevel.aidl). Selecting a level
+     * different from SpatializationLevel.NONE with create the specialized multichannel output
+     * mixer, create and enable the spatializer effect and let the audio policy attach eligible
+     * AudioTrack to this output stream.
+     */
+    void setLevel(SpatializationLevel level);
+
+    /** Gets the selected spatialization level (see SpatializationLevel.aidl) */
+    SpatializationLevel getLevel();
+
+    /** Reports if the spatializer engine supports head tracking or not.
+     * This is a pre condition independent of the fact that a head tracking sensor is
+     * registered or not.
+     */
+    boolean isHeadTrackingSupported();
+
+    /** Reports the list of supported head tracking modes (see SpatializerHeadTrackingMode.aidl).
+     * The list can be empty if the spatializer implementation does not support head tracking or if
+     * no head tracking sensor is registered (see setHeadSensor() and setScreenSensor()).
+     */
+    SpatializerHeadTrackingMode[] getSupportedHeadTrackingModes();
+
+    /** Selects the desired head tracking mode (see SpatializerHeadTrackingMode.aidl) */
+    void setDesiredHeadTrackingMode(SpatializerHeadTrackingMode mode);
+
+    /** Gets the actual head tracking mode. Can be different from the desired mode if conditions to
+     * enable the desired mode are not met (e.g if the head tracking device was removed)
+     */
+    SpatializerHeadTrackingMode getActualHeadTrackingMode();
+
+    /** Reset the head tracking algorithm to consider current head pose as neutral */
+    void recenterHeadTracker();
+
+    /** Set the screen to stage transform to use by the head tracking algorithm
+     * The screen to stage transform is conveyed as a vector of 6 elements,
+     * where the first three are a translation vector and
+     * the last three are a rotation vector.
+     */
+    void setGlobalTransform(in float[] screenToStage);
+
+    /**
+     * Set the sensor that is to be used for head-tracking.
+     * -1 can be used to disable head-tracking.
+     */
+    void setHeadSensor(int sensorHandle);
+
+    /**
+     * Set the sensor that is to be used for screen-tracking.
+     * -1 can be used to disable screen-tracking.
+     */
+    void setScreenSensor(int sensorHandle);
+
+    /**
+     * Sets the display orientation.
+     * Orientation is expressed in the angle of rotation from the physical "up" side of the screen
+     * to the logical "up" side of the content displayed the screen. Counterclockwise angles, as
+     * viewed while facing the screen are positive.
+     */
+    void setDisplayOrientation(float physicalToLogicalAngle);
+
+    /**
+     * Sets the hinge angle for foldable devices.
+     */
+    void setHingeAngle(float hingeAngle);
+
+    /** Reports the list of supported spatialization modess (see SpatializationMode.aidl).
+     * The list should never be empty if an ISpatializer interface was successfully
+     * retrieved with IAudioPolicyService.getSpatializer().
+     */
+    SpatializationMode[] getSupportedModes();
+
+    /**
+     * Registers a callback to monitor head tracking functions.
+     * Only one callback can be registered on a Spatializer.
+     * The last callback registered wins and passing a nullptr unregisters
+     * last registered callback.
+     */
+    void registerHeadTrackingCallback(@nullable ISpatializerHeadTrackingCallback callback);
+
+    /**
+     * Sets a parameter to the spatializer engine. Used by effect implementor for vendor
+     * specific configuration.
+     */
+     void setParameter(int key, in byte[] value);
+
+    /**
+     * Gets a parameter from the spatializer engine. Used by effect implementor for vendor
+     * specific configuration.
+     */
+     void getParameter(int key, inout byte[] value);
+
+    /**
+     * Gets the io handle of the output stream the spatializer is connected to.
+     */
+     int getOutput();
+}
diff --git a/media/libaudioclient/aidl/android/media/ISpatializerHeadTrackingCallback.aidl b/media/libaudioclient/aidl/android/media/ISpatializerHeadTrackingCallback.aidl
new file mode 100644
index 0000000..23d5e13
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/ISpatializerHeadTrackingCallback.aidl
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+import android.media.SpatializationLevel;
+import android.media.SpatializerHeadTrackingMode;
+
+/**
+ * The ISpatializerHeadTrackingCallback interface is a callback associated to the
+ * Spatializer head tracking function. It can be registered via the ISpatializer
+ * interface to monitor head tracking related states.
+ * {@hide}
+ */
+oneway interface ISpatializerHeadTrackingCallback {
+    /** Called when the head tracking mode has changed
+     */
+    void onHeadTrackingModeChanged(SpatializerHeadTrackingMode mode);
+
+    /** Called when the head to stage pose hase been updated
+     * The head to stage pose is conveyed as a vector of 6 elements,
+     * where the first three are a translation vector and
+     * the last three are a rotation vector.
+     */
+    void onHeadToSoundStagePoseUpdated(in float[] headToStage);
+}
diff --git a/media/libaudioclient/aidl/android/media/OpenOutputRequest.aidl b/media/libaudioclient/aidl/android/media/OpenOutputRequest.aidl
index 06b12e9..1541948 100644
--- a/media/libaudioclient/aidl/android/media/OpenOutputRequest.aidl
+++ b/media/libaudioclient/aidl/android/media/OpenOutputRequest.aidl
@@ -17,6 +17,7 @@
 package android.media;
 
 import android.media.AudioConfig;
+import android.media.AudioConfigBase;
 import android.media.AudioPort;
 
 /**
@@ -25,7 +26,8 @@
 parcelable OpenOutputRequest {
     /** Interpreted as audio_module_handle_t. */
     int module;
-    AudioConfig config;
+    AudioConfig halConfig;
+    AudioConfigBase mixerConfig;
     /** Type must be DEVICE. */
     AudioPort device;
     /** Bitmask, indexed by AudioOutputFlag. */
diff --git a/media/libaudioclient/aidl/android/media/SpatializationLevel.aidl b/media/libaudioclient/aidl/android/media/SpatializationLevel.aidl
new file mode 100644
index 0000000..961c5a1
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/SpatializationLevel.aidl
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+/**
+ * The spatialization level supported by the spatializer stage effect implementation.
+ * Used by methods of the ISpatializer interface.
+ * {@hide}
+ */
+@Backing(type="byte")
+enum SpatializationLevel {
+    /** Spatialization is disabled. */
+    NONE = 0,
+    /** The spatializer accepts audio with positional multichannel masks (e.g 5.1). */
+    SPATIALIZER_MULTICHANNEL = 1,
+    /** The spatializer accepts audio made of a channel bed of positional multichannels (e.g 5.1)
+     * and audio objects positioned independently via meta data.
+     */
+    SPATIALIZER_MCHAN_BED_PLUS_OBJECTS = 2,
+}
diff --git a/media/libaudioclient/aidl/android/media/SpatializationMode.aidl b/media/libaudioclient/aidl/android/media/SpatializationMode.aidl
new file mode 100644
index 0000000..5d8fd93
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/SpatializationMode.aidl
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+/**
+ * The spatialization mode supported by the spatializer stage effect implementation.
+ * Used by methods of the ISpatializer interface.
+ * {@hide}
+ */
+@Backing(type="byte")
+enum SpatializationMode {
+    /** The spatializer supports binaural mode (over headphones type devices). */
+    SPATIALIZATER_BINAURAL = 0,
+    /** The spatializer supports transaural mode (over speaker type devices). */
+    SPATIALIZATER_TRANSAURAL = 1,
+}
diff --git a/media/libaudioclient/aidl/android/media/SpatializerHeadTrackingMode.aidl b/media/libaudioclient/aidl/android/media/SpatializerHeadTrackingMode.aidl
new file mode 100644
index 0000000..58e0f61
--- /dev/null
+++ b/media/libaudioclient/aidl/android/media/SpatializerHeadTrackingMode.aidl
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.media;
+
+
+/**
+ * The head tracking mode supported by the spatializer effect implementation.
+ * Used by methods of the ISpatializer interface.
+ * {@hide}
+ */
+@Backing(type="byte")
+enum SpatializerHeadTrackingMode {
+    /** Head tracking is active in a mode not listed below (forward compatibility) */
+    OTHER = 0,
+    /** Head tracking is disabled */
+    DISABLED = 1,
+    /** Head tracking is performed relative to the real work environment */
+    RELATIVE_WORLD = 2,
+    /** Head tracking is performed relative to the device's screen */
+    RELATIVE_SCREEN = 3,
+}
diff --git a/media/libaudioclient/fuzzer/audioflinger_fuzzer.cpp b/media/libaudioclient/fuzzer/audioflinger_fuzzer.cpp
index d03c6fa..bd9e158 100644
--- a/media/libaudioclient/fuzzer/audioflinger_fuzzer.cpp
+++ b/media/libaudioclient/fuzzer/audioflinger_fuzzer.cpp
@@ -383,6 +383,9 @@
                                    const std::vector<uint8_t> &replyData __unused) override {
         return binder::Status::ok();
     }
+    binder::Status framesProcessed(int32_t frames __unused) override {
+        return binder::Status::ok();
+    }
 };
 
 status_t AudioFlingerFuzzer::invokeAudioEffect() {
@@ -424,6 +427,7 @@
     request.attributionSource.packageName = opPackageName;
     request.attributionSource.pid = VALUE_OR_RETURN_STATUS(legacy2aidl_pid_t_int32_t(getpid()));
     request.probe = false;
+    request.notifyFramesProcessed = false;
 
     media::CreateEffectResponse response{};
     status_t status = af->createEffect(request, &response);
@@ -648,11 +652,15 @@
     sp<DeviceDescriptorBase> device = new DeviceDescriptorBase(getValue(&mFdp, kDevices));
     audio_output_flags_t flags = getValue(&mFdp, kOutputFlags);
 
+    audio_config_base_t mixerConfig = AUDIO_CONFIG_BASE_INITIALIZER;
+
     media::OpenOutputRequest request{};
     media::OpenOutputResponse response{};
 
     request.module = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_module_handle_t_int32_t(module));
-    request.config = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_t_AudioConfig(config));
+    request.halConfig = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_t_AudioConfig(config));
+    request.mixerConfig =
+            VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_base_t_AudioConfigBase(mixerConfig));
     request.device = VALUE_OR_RETURN_STATUS(legacy2aidl_DeviceDescriptorBase(device));
     request.flags = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_output_flags_t_int32_t_mask(flags));
 
diff --git a/media/libaudioclient/include/media/AudioCommonTypes.h b/media/libaudioclient/include/media/AudioCommonTypes.h
index 5dfe5fc..5f0c590 100644
--- a/media/libaudioclient/include/media/AudioCommonTypes.h
+++ b/media/libaudioclient/include/media/AudioCommonTypes.h
@@ -41,6 +41,42 @@
     return !(lhs==rhs);
 }
 
+constexpr bool operator==(const audio_offload_info_t &lhs, const audio_offload_info_t &rhs)
+{
+    return lhs.version == rhs.version && lhs.size == rhs.size &&
+           lhs.sample_rate == rhs.sample_rate && lhs.channel_mask == rhs.channel_mask &&
+           lhs.format == rhs.format && lhs.stream_type == rhs.stream_type &&
+           lhs.bit_rate == rhs.bit_rate && lhs.duration_us == rhs.duration_us &&
+           lhs.has_video == rhs.has_video && lhs.is_streaming == rhs.is_streaming &&
+           lhs.bit_width == rhs.bit_width && lhs.offload_buffer_size == rhs.offload_buffer_size &&
+           lhs.usage == rhs.usage && lhs.encapsulation_mode == rhs.encapsulation_mode &&
+           lhs.content_id == rhs.content_id && lhs.sync_id == rhs.sync_id;
+}
+constexpr bool operator!=(const audio_offload_info_t &lhs, const audio_offload_info_t &rhs)
+{
+    return !(lhs==rhs);
+}
+
+constexpr bool operator==(const audio_config_t &lhs, const audio_config_t &rhs)
+{
+    return lhs.sample_rate == rhs.sample_rate && lhs.channel_mask == rhs.channel_mask &&
+           lhs.format == rhs.format && lhs.offload_info == rhs.offload_info;
+}
+constexpr bool operator!=(const audio_config_t &lhs, const audio_config_t &rhs)
+{
+    return !(lhs==rhs);
+}
+
+constexpr bool operator==(const audio_config_base_t &lhs, const audio_config_base_t &rhs)
+{
+    return lhs.sample_rate == rhs.sample_rate && lhs.channel_mask == rhs.channel_mask &&
+           lhs.format == rhs.format;
+}
+constexpr bool operator!=(const audio_config_base_t &lhs, const audio_config_base_t &rhs)
+{
+    return !(lhs==rhs);
+}
+
 enum volume_group_t : uint32_t;
 static const volume_group_t VOLUME_GROUP_NONE = static_cast<volume_group_t>(-1);
 
diff --git a/media/libaudioclient/include/media/AudioEffect.h b/media/libaudioclient/include/media/AudioEffect.h
index 3c19ec1..dd4d2da 100644
--- a/media/libaudioclient/include/media/AudioEffect.h
+++ b/media/libaudioclient/include/media/AudioEffect.h
@@ -283,7 +283,8 @@
         EVENT_CONTROL_STATUS_CHANGED = 0,
         EVENT_ENABLE_STATUS_CHANGED = 1,
         EVENT_PARAMETER_CHANGED = 2,
-        EVENT_ERROR = 3
+        EVENT_ERROR = 3,
+        EVENT_FRAMES_PROCESSED = 4,
     };
 
     /* Callback function notifying client application of a change in effect engine state or
@@ -389,7 +390,8 @@
                             audio_session_t sessionId = AUDIO_SESSION_OUTPUT_MIX,
                             audio_io_handle_t io = AUDIO_IO_HANDLE_NONE,
                             const AudioDeviceTypeAddr& device = {},
-                            bool probe = false);
+                            bool probe = false,
+                            bool notifyFramesProcessed = false);
     /*
      * Same as above but with type and uuid specified by character strings.
      */
@@ -401,7 +403,8 @@
                             audio_session_t sessionId = AUDIO_SESSION_OUTPUT_MIX,
                             audio_io_handle_t io = AUDIO_IO_HANDLE_NONE,
                             const AudioDeviceTypeAddr& device = {},
-                            bool probe = false);
+                            bool probe = false,
+                            bool notifyFramesProcessed = false);
 
     /* Result of constructing the AudioEffect. This must be checked
      * before using any AudioEffect API.
@@ -552,6 +555,7 @@
      virtual void commandExecuted(int32_t cmdCode,
                                   const std::vector<uint8_t>& cmdData,
                                   const std::vector<uint8_t>& replyData);
+     virtual void framesProcessed(int32_t frames);
 
 private:
 
@@ -587,6 +591,14 @@
             }
             return binder::Status::ok();
         }
+        binder::Status framesProcessed(int32_t frames) override {
+            sp<AudioEffect> effect = mEffect.promote();
+            if (effect != 0) {
+                effect->framesProcessed(frames);
+            }
+            return binder::Status::ok();
+        }
+
 
         // IBinder::DeathRecipient
         virtual void binderDied(const wp<IBinder>& /*who*/) {
diff --git a/media/libaudioclient/include/media/AudioRecord.h b/media/libaudioclient/include/media/AudioRecord.h
index 326919a..f17ee3a 100644
--- a/media/libaudioclient/include/media/AudioRecord.h
+++ b/media/libaudioclient/include/media/AudioRecord.h
@@ -264,6 +264,7 @@
             size_t      frameCount() const  { return mFrameCount; }
             size_t      frameSize() const   { return mFrameSize; }
             audio_source_t inputSource() const  { return mAttributes.source; }
+            audio_channel_mask_t channelMask() const { return mChannelMask; }
 
     /*
      * Return the period of the notification callback in frames.
diff --git a/media/libaudioclient/include/media/AudioSystem.h b/media/libaudioclient/include/media/AudioSystem.h
index 8ba23ad..869bd6e 100644
--- a/media/libaudioclient/include/media/AudioSystem.h
+++ b/media/libaudioclient/include/media/AudioSystem.h
@@ -22,6 +22,8 @@
 #include <android/media/AudioVibratorInfo.h>
 #include <android/media/BnAudioFlingerClient.h>
 #include <android/media/BnAudioPolicyServiceClient.h>
+#include <android/media/INativeSpatializerCallback.h>
+#include <android/media/ISpatializer.h>
 #include <android/content/AttributionSourceState.h>
 #include <media/AidlConversionUtil.h>
 #include <media/AudioDeviceTypeAddr.h>
@@ -225,6 +227,9 @@
     // Indicate JAVA services are ready (scheduling, power management ...)
     static status_t systemReady();
 
+    // Indicate audio policy service is ready
+    static status_t audioPolicyReady();
+
     // Returns the number of frames per audio HAL buffer.
     // Corresponds to audio_stream->get_buffer_size()/audio_stream_in_frame_size() for input.
     // See also getFrameCount().
@@ -485,6 +490,49 @@
     static status_t getDeviceForStrategy(product_strategy_t strategy,
             AudioDeviceTypeAddr &device);
 
+
+    /**
+     * If a spatializer stage effect is present on the platform, this will return an
+     * ISpatializer interface to control this feature.
+     * If no spatializer stage is present, a null interface is returned.
+     * The INativeSpatializerCallback passed must not be null.
+     * Only one ISpatializer interface can exist at a given time. The native audio policy
+     * service will reject the request if an interface was already acquired and previous owner
+     * did not die or call ISpatializer.release().
+     * @param callback in: the callback to receive state updates if the ISpatializer
+     *        interface is acquired.
+     * @param spatializer out: the ISpatializer interface made available to control the
+     *        platform spatializer
+     * @return NO_ERROR in case of success, DEAD_OBJECT, NO_INIT, PERMISSION_DENIED, BAD_VALUE
+     *         in case of error.
+     */
+    static status_t getSpatializer(const sp<media::INativeSpatializerCallback>& callback,
+                                        sp<media::ISpatializer>* spatializer);
+
+    /**
+     * Queries if some kind of spatialization will be performed if the audio playback context
+     * described by the provided arguments is present.
+     * The context is made of:
+     * - The audio attributes describing the playback use case.
+     * - The audio configuration describing the audio format, channels, sampling rate ...
+     * - The devices describing the sink audio device selected for playback.
+     * All arguments are optional and only the specified arguments are used to match against
+     * supported criteria. For instance, supplying no argument will tell if spatialization is
+     * supported or not in general.
+     * @param attr audio attributes describing the playback use case
+     * @param config audio configuration describing the audio format, channels, sampling rate...
+     * @param devices the sink audio device selected for playback
+     * @param canBeSpatialized out: true if spatialization is enabled for this context,
+     *        false otherwise
+     * @return NO_ERROR in case of success, DEAD_OBJECT, NO_INIT, BAD_VALUE
+     *         in case of error.
+     */
+    static status_t canBeSpatialized(const audio_attributes_t *attr,
+                                     const audio_config_t *config,
+                                     const AudioDeviceTypeAddrVector &devices,
+                                     bool *canBeSpatialized);
+
+
     // A listener for capture state changes.
     class CaptureStateListener : public RefBase {
     public:
@@ -497,11 +545,11 @@
         virtual ~CaptureStateListener() = default;
     };
 
-    // Regiseters a listener for sound trigger capture state changes.
+    // Registers a listener for sound trigger capture state changes.
     // There may only be one such listener registered at any point.
-    // The listener onStateChanged() method will be invoked sychronously from
+    // The listener onStateChanged() method will be invoked synchronously from
     // this call with the initial value.
-    // The listener onServiceDied() method will be invoked sychronously from
+    // The listener onServiceDied() method will be invoked synchronously from
     // this call if initial attempt to register failed.
     // If the audio policy service cannot be reached, this method will return
     // PERMISSION_DENIED and will not invoke the callback, otherwise, it will
diff --git a/media/libaudioclient/include/media/AudioTrack.h b/media/libaudioclient/include/media/AudioTrack.h
index cb00990..5fea637 100644
--- a/media/libaudioclient/include/media/AudioTrack.h
+++ b/media/libaudioclient/include/media/AudioTrack.h
@@ -401,6 +401,7 @@
 
             uint32_t    channelCount() const { return mChannelCount; }
             size_t      frameCount() const  { return mFrameCount; }
+            audio_channel_mask_t channelMask() const { return mChannelMask; }
 
     /*
      * Return the period of the notification callback in frames.
diff --git a/media/libaudioclient/include/media/IAudioFlinger.h b/media/libaudioclient/include/media/IAudioFlinger.h
index 0e059f7..9e5019e 100644
--- a/media/libaudioclient/include/media/IAudioFlinger.h
+++ b/media/libaudioclient/include/media/IAudioFlinger.h
@@ -329,6 +329,9 @@
     /* Indicate JAVA services are ready (scheduling, power management ...) */
     virtual status_t systemReady() = 0;
 
+    // Indicate audio policy service is ready
+    virtual status_t audioPolicyReady() = 0;
+
     // Returns the number of frames per audio HAL buffer.
     virtual size_t frameCountHAL(audio_io_handle_t ioHandle) const = 0;
 
@@ -432,6 +435,8 @@
     status_t setAudioPortConfig(const struct audio_port_config* config) override;
     audio_hw_sync_t getAudioHwSyncForSession(audio_session_t sessionId) override;
     status_t systemReady() override;
+    status_t audioPolicyReady() override;
+
     size_t frameCountHAL(audio_io_handle_t ioHandle) const override;
     status_t getMicrophones(std::vector<media::MicrophoneInfo>* microphones) override;
     status_t setAudioHalPids(const std::vector<pid_t>& pids) override;
@@ -514,6 +519,7 @@
             SET_AUDIO_PORT_CONFIG = media::BnAudioFlingerService::TRANSACTION_setAudioPortConfig,
             GET_AUDIO_HW_SYNC_FOR_SESSION = media::BnAudioFlingerService::TRANSACTION_getAudioHwSyncForSession,
             SYSTEM_READY = media::BnAudioFlingerService::TRANSACTION_systemReady,
+            AUDIO_POLICY_READY = media::BnAudioFlingerService::TRANSACTION_audioPolicyReady,
             FRAME_COUNT_HAL = media::BnAudioFlingerService::TRANSACTION_frameCountHAL,
             GET_MICROPHONES = media::BnAudioFlingerService::TRANSACTION_getMicrophones,
             SET_MASTER_BALANCE = media::BnAudioFlingerService::TRANSACTION_setMasterBalance,
@@ -624,6 +630,7 @@
     Status setAudioPortConfig(const media::AudioPortConfig& config) override;
     Status getAudioHwSyncForSession(int32_t sessionId, int32_t* _aidl_return) override;
     Status systemReady() override;
+    Status audioPolicyReady() override;
     Status frameCountHAL(int32_t ioHandle, int64_t* _aidl_return) override;
     Status getMicrophones(std::vector<media::MicrophoneInfoData>* _aidl_return) override;
     Status setAudioHalPids(const std::vector<int32_t>& pids) override;
diff --git a/media/libaudiohal/impl/ConversionHelperHidl.cpp b/media/libaudiohal/impl/ConversionHelperHidl.cpp
index 32eaa31..0503698 100644
--- a/media/libaudiohal/impl/ConversionHelperHidl.cpp
+++ b/media/libaudiohal/impl/ConversionHelperHidl.cpp
@@ -105,6 +105,15 @@
 }
 
 // static
+void ConversionHelperHidl::argsFromHal(
+        const Vector<String16>& args, hidl_vec<hidl_string> *hidlArgs) {
+    hidlArgs->resize(args.size());
+    for (size_t i = 0; i < args.size(); ++i) {
+        (*hidlArgs)[i] = String8(args[i]).c_str();
+    }
+}
+
+// static
 status_t ConversionHelperHidl::analyzeResult(const Result& result) {
     switch (result) {
         case Result::OK: return OK;
diff --git a/media/libaudiohal/impl/ConversionHelperHidl.h b/media/libaudiohal/impl/ConversionHelperHidl.h
index 59122c7..2909013 100644
--- a/media/libaudiohal/impl/ConversionHelperHidl.h
+++ b/media/libaudiohal/impl/ConversionHelperHidl.h
@@ -21,6 +21,8 @@
 #include <hidl/HidlSupport.h>
 #include <system/audio.h>
 #include <utils/String8.h>
+#include <utils/String16.h>
+#include <utils/Vector.h>
 
 using ::android::hardware::audio::CPP_VERSION::ParameterValue;
 using CoreResult = ::android::hardware::audio::CPP_VERSION::Result;
@@ -37,6 +39,7 @@
     static status_t keysFromHal(const String8& keys, hidl_vec<hidl_string> *hidlKeys);
     static status_t parametersFromHal(const String8& kvPairs, hidl_vec<ParameterValue> *hidlParams);
     static void parametersToHal(const hidl_vec<ParameterValue>& parameters, String8 *values);
+    static void argsFromHal(const Vector<String16>& args, hidl_vec<hidl_string> *hidlArgs);
 
     ConversionHelperHidl(const char* className);
 
diff --git a/media/libaudiohal/impl/DeviceHalHidl.cpp b/media/libaudiohal/impl/DeviceHalHidl.cpp
index 02d66ae..aa94eea 100644
--- a/media/libaudiohal/impl/DeviceHalHidl.cpp
+++ b/media/libaudiohal/impl/DeviceHalHidl.cpp
@@ -457,11 +457,13 @@
 }
 #endif
 
-status_t DeviceHalHidl::dump(int fd) {
+status_t DeviceHalHidl::dump(int fd, const Vector<String16>& args) {
     if (mDevice == 0) return NO_INIT;
     native_handle_t* hidlHandle = native_handle_create(1, 0);
     hidlHandle->data[0] = fd;
-    Return<void> ret = mDevice->debug(hidlHandle, {} /* options */);
+    hidl_vec<hidl_string> hidlArgs;
+    argsFromHal(args, &hidlArgs);
+    Return<void> ret = mDevice->debug(hidlHandle, hidlArgs);
     native_handle_delete(hidlHandle);
 
     // TODO(b/111997867, b/177271958)  Workaround - remove when fixed.
diff --git a/media/libaudiohal/impl/DeviceHalHidl.h b/media/libaudiohal/impl/DeviceHalHidl.h
index 2c847cf..2694ab3 100644
--- a/media/libaudiohal/impl/DeviceHalHidl.h
+++ b/media/libaudiohal/impl/DeviceHalHidl.h
@@ -119,7 +119,7 @@
     status_t addDeviceEffect(audio_port_handle_t device, sp<EffectHalInterface> effect) override;
     status_t removeDeviceEffect(audio_port_handle_t device, sp<EffectHalInterface> effect) override;
 
-    virtual status_t dump(int fd);
+    status_t dump(int fd, const Vector<String16>& args) override;
 
   private:
     friend class DevicesFactoryHalHidl;
diff --git a/media/libaudiohal/impl/DeviceHalLocal.cpp b/media/libaudiohal/impl/DeviceHalLocal.cpp
index af7dc1a..e0304af 100644
--- a/media/libaudiohal/impl/DeviceHalLocal.cpp
+++ b/media/libaudiohal/impl/DeviceHalLocal.cpp
@@ -233,7 +233,7 @@
     return INVALID_OPERATION;
 }
 
-status_t DeviceHalLocal::dump(int fd) {
+status_t DeviceHalLocal::dump(int fd, const Vector<String16>& /* args */) {
     return mDev->dump(mDev, fd);
 }
 
diff --git a/media/libaudiohal/impl/DeviceHalLocal.h b/media/libaudiohal/impl/DeviceHalLocal.h
index 46b510b..2fde936 100644
--- a/media/libaudiohal/impl/DeviceHalLocal.h
+++ b/media/libaudiohal/impl/DeviceHalLocal.h
@@ -112,7 +112,7 @@
     status_t addDeviceEffect(audio_port_handle_t device, sp<EffectHalInterface> effect) override;
     status_t removeDeviceEffect(audio_port_handle_t device, sp<EffectHalInterface> effect) override;
 
-    virtual status_t dump(int fd);
+    status_t dump(int fd, const Vector<String16>& args) override;
 
     void closeOutputStream(struct audio_stream_out *stream_out);
     void closeInputStream(struct audio_stream_in *stream_in);
diff --git a/media/libaudiohal/impl/EffectsFactoryHalHidl.cpp b/media/libaudiohal/impl/EffectsFactoryHalHidl.cpp
index f042b92..ffe0d72 100644
--- a/media/libaudiohal/impl/EffectsFactoryHalHidl.cpp
+++ b/media/libaudiohal/impl/EffectsFactoryHalHidl.cpp
@@ -73,7 +73,9 @@
         uint32_t index, effect_descriptor_t *pDescriptor) {
     // TODO: We need somehow to track the changes on the server side
     // or figure out how to convert everybody to query all the descriptors at once.
-    // TODO: check for nullptr
+    if (pDescriptor == nullptr) {
+        return BAD_VALUE;
+    }
     if (mLastDescriptors.size() == 0) {
         status_t queryResult = queryAllDescriptors();
         if (queryResult != OK) return queryResult;
@@ -85,7 +87,9 @@
 
 status_t EffectsFactoryHalHidl::getDescriptor(
         const effect_uuid_t *pEffectUuid, effect_descriptor_t *pDescriptor) {
-    // TODO: check for nullptr
+    if (pDescriptor == nullptr || pEffectUuid == nullptr) {
+        return BAD_VALUE;
+    }
     if (mEffectsFactory == 0) return NO_INIT;
     Uuid hidlUuid;
     UuidUtils::uuidFromHal(*pEffectUuid, &hidlUuid);
@@ -105,6 +109,33 @@
     return processReturn(__FUNCTION__, ret);
 }
 
+status_t EffectsFactoryHalHidl::getDescriptors(const effect_uuid_t *pEffectType,
+                                               std::vector<effect_descriptor_t> *descriptors) {
+    if (pEffectType == nullptr || descriptors == nullptr) {
+        return BAD_VALUE;
+    }
+
+    uint32_t numEffects = 0;
+    status_t status = queryNumberEffects(&numEffects);
+    if (status != NO_ERROR) {
+        ALOGW("%s error %d from FactoryHal queryNumberEffects", __func__, status);
+        return status;
+    }
+
+    for (uint32_t i = 0; i < numEffects; i++) {
+        effect_descriptor_t descriptor;
+        status = getDescriptor(i, &descriptor);
+        if (status != NO_ERROR) {
+            ALOGW("%s error %d from FactoryHal getDescriptor", __func__, status);
+            continue;
+        }
+        if (memcmp(&descriptor.type, pEffectType, sizeof(effect_uuid_t)) == 0) {
+            descriptors->push_back(descriptor);
+        }
+    }
+    return descriptors->empty() ? NAME_NOT_FOUND : NO_ERROR;
+}
+
 status_t EffectsFactoryHalHidl::createEffect(
         const effect_uuid_t *pEffectUuid, int32_t sessionId, int32_t ioId,
         int32_t deviceId __unused, sp<EffectHalInterface> *effect) {
diff --git a/media/libaudiohal/impl/EffectsFactoryHalHidl.h b/media/libaudiohal/impl/EffectsFactoryHalHidl.h
index 5fa85e7..ff26d9f 100644
--- a/media/libaudiohal/impl/EffectsFactoryHalHidl.h
+++ b/media/libaudiohal/impl/EffectsFactoryHalHidl.h
@@ -45,6 +45,9 @@
     virtual status_t getDescriptor(const effect_uuid_t *pEffectUuid,
             effect_descriptor_t *pDescriptor);
 
+    virtual status_t getDescriptors(const effect_uuid_t *pEffectType,
+                                    std::vector<effect_descriptor_t> *descriptors);
+
     // Creates an effect engine of the specified type.
     // To release the effect engine, it is necessary to release references
     // to the returned effect object.
diff --git a/media/libaudiohal/impl/StreamHalHidl.cpp b/media/libaudiohal/impl/StreamHalHidl.cpp
index 129b1c1..e63aded 100644
--- a/media/libaudiohal/impl/StreamHalHidl.cpp
+++ b/media/libaudiohal/impl/StreamHalHidl.cpp
@@ -152,11 +152,13 @@
     return processReturn("standby", mStream->standby());
 }
 
-status_t StreamHalHidl::dump(int fd) {
+status_t StreamHalHidl::dump(int fd, const Vector<String16>& args) {
     if (!mStream) return NO_INIT;
     native_handle_t* hidlHandle = native_handle_create(1, 0);
     hidlHandle->data[0] = fd;
-    Return<void> ret = mStream->debug(hidlHandle, {} /* options */);
+    hidl_vec<hidl_string> hidlArgs;
+    argsFromHal(args, &hidlArgs);
+    Return<void> ret = mStream->debug(hidlHandle, hidlArgs);
     native_handle_delete(hidlHandle);
 
     // TODO(b/111997867, b/177271958)  Workaround - remove when fixed.
diff --git a/media/libaudiohal/impl/StreamHalHidl.h b/media/libaudiohal/impl/StreamHalHidl.h
index 970903b..6f5dd04 100644
--- a/media/libaudiohal/impl/StreamHalHidl.h
+++ b/media/libaudiohal/impl/StreamHalHidl.h
@@ -71,7 +71,7 @@
     // Put the audio hardware input/output into standby mode.
     virtual status_t standby();
 
-    virtual status_t dump(int fd);
+    virtual status_t dump(int fd, const Vector<String16>& args) override;
 
     // Start a stream operating in mmap mode.
     virtual status_t start();
diff --git a/media/libaudiohal/impl/StreamHalLocal.cpp b/media/libaudiohal/impl/StreamHalLocal.cpp
index 34bd5df..11fac61 100644
--- a/media/libaudiohal/impl/StreamHalLocal.cpp
+++ b/media/libaudiohal/impl/StreamHalLocal.cpp
@@ -87,7 +87,8 @@
     return mStream->standby(mStream);
 }
 
-status_t StreamHalLocal::dump(int fd) {
+status_t StreamHalLocal::dump(int fd, const Vector<String16>& args) {
+    (void) args;
     status_t status = mStream->dump(mStream, fd);
     mStreamPowerLog.dump(fd);
     return status;
diff --git a/media/libaudiohal/impl/StreamHalLocal.h b/media/libaudiohal/impl/StreamHalLocal.h
index b260495..493c521 100644
--- a/media/libaudiohal/impl/StreamHalLocal.h
+++ b/media/libaudiohal/impl/StreamHalLocal.h
@@ -50,7 +50,7 @@
     // Put the audio hardware input/output into standby mode.
     virtual status_t standby();
 
-    virtual status_t dump(int fd);
+    virtual status_t dump(int fd, const Vector<String16>& args) override;
 
     // Start a stream operating in mmap mode.
     virtual status_t start() = 0;
diff --git a/media/libaudiohal/include/media/audiohal/DeviceHalInterface.h b/media/libaudiohal/include/media/audiohal/DeviceHalInterface.h
index 29ef011..69cbcec 100644
--- a/media/libaudiohal/include/media/audiohal/DeviceHalInterface.h
+++ b/media/libaudiohal/include/media/audiohal/DeviceHalInterface.h
@@ -120,7 +120,7 @@
     virtual status_t removeDeviceEffect(
             audio_port_handle_t device, sp<EffectHalInterface> effect) = 0;
 
-    virtual status_t dump(int fd) = 0;
+    virtual status_t dump(int fd, const Vector<String16>& args) = 0;
 
   protected:
     // Subclasses can not be constructed directly by clients.
diff --git a/media/libaudiohal/include/media/audiohal/EffectsFactoryHalInterface.h b/media/libaudiohal/include/media/audiohal/EffectsFactoryHalInterface.h
index 9fb56ae..3e505bd 100644
--- a/media/libaudiohal/include/media/audiohal/EffectsFactoryHalInterface.h
+++ b/media/libaudiohal/include/media/audiohal/EffectsFactoryHalInterface.h
@@ -37,6 +37,9 @@
     virtual status_t getDescriptor(const effect_uuid_t *pEffectUuid,
             effect_descriptor_t *pDescriptor) = 0;
 
+    virtual status_t getDescriptors(const effect_uuid_t *pEffectType,
+                                    std::vector<effect_descriptor_t> *descriptors) = 0;
+
     // Creates an effect engine of the specified type.
     // To release the effect engine, it is necessary to release references
     // to the returned effect object.
diff --git a/media/libaudiohal/include/media/audiohal/StreamHalInterface.h b/media/libaudiohal/include/media/audiohal/StreamHalInterface.h
index 2be12fb..2b5b2db 100644
--- a/media/libaudiohal/include/media/audiohal/StreamHalInterface.h
+++ b/media/libaudiohal/include/media/audiohal/StreamHalInterface.h
@@ -25,6 +25,7 @@
 #include <utils/Errors.h>
 #include <utils/RefBase.h>
 #include <utils/String8.h>
+#include <utils/Vector.h>
 
 namespace android {
 
@@ -69,7 +70,7 @@
     // Put the audio hardware input/output into standby mode.
     virtual status_t standby() = 0;
 
-    virtual status_t dump(int fd) = 0;
+    virtual status_t dump(int fd, const Vector<String16>& args = {}) = 0;
 
     // Start a stream operating in mmap mode.
     virtual status_t start() = 0;
diff --git a/media/libaudioprocessing/AudioMixer.cpp b/media/libaudioprocessing/AudioMixer.cpp
index d85e2e9..e68c002 100644
--- a/media/libaudioprocessing/AudioMixer.cpp
+++ b/media/libaudioprocessing/AudioMixer.cpp
@@ -434,6 +434,12 @@
                 track->mHapticIntensity = hapticIntensity;
             }
             } break;
+        case HAPTIC_MAX_AMPLITUDE: {
+            const float hapticMaxAmplitude = *reinterpret_cast<float*>(value);
+            if (track->mHapticMaxAmplitude != hapticMaxAmplitude) {
+                track->mHapticMaxAmplitude = hapticMaxAmplitude;
+            }
+            } break;
         default:
             LOG_ALWAYS_FATAL("setParameter track: bad param %d", param);
         }
@@ -553,6 +559,7 @@
     // haptic
     t->mHapticPlaybackEnabled = false;
     t->mHapticIntensity = os::HapticScale::NONE;
+    t->mHapticMaxAmplitude = NAN;
     t->mMixerHapticChannelMask = AUDIO_CHANNEL_NONE;
     t->mMixerHapticChannelCount = 0;
     t->mAdjustInChannelCount = t->channelCount + t->mHapticChannelCount;
@@ -602,7 +609,8 @@
                 switch (t->mMixerFormat) {
                 // Mixer format should be AUDIO_FORMAT_PCM_FLOAT.
                 case AUDIO_FORMAT_PCM_FLOAT: {
-                    os::scaleHapticData((float*) buffer, sampleCount, t->mHapticIntensity);
+                    os::scaleHapticData((float*) buffer, sampleCount, t->mHapticIntensity,
+                                        t->mHapticMaxAmplitude);
                 } break;
                 default:
                     LOG_ALWAYS_FATAL("bad mMixerFormat: %#x", t->mMixerFormat);
diff --git a/media/libaudioprocessing/AudioMixerOps.h b/media/libaudioprocessing/AudioMixerOps.h
index cd47dc6..ab6a8b6 100644
--- a/media/libaudioprocessing/AudioMixerOps.h
+++ b/media/libaudioprocessing/AudioMixerOps.h
@@ -17,6 +17,8 @@
 #ifndef ANDROID_AUDIO_MIXER_OPS_H
 #define ANDROID_AUDIO_MIXER_OPS_H
 
+#include <audio_utils/channels.h>
+#include <audio_utils/primitives.h>
 #include <system/audio.h>
 
 namespace android {
@@ -229,15 +231,26 @@
  * complexity of working on interleaved streams is now getting
  * too high, and likely limits compiler optimization.
  */
-template <int MIXTYPE, int NCHAN,
+
+// compile-time function.
+constexpr inline bool usesCenterChannel(audio_channel_mask_t mask) {
+    using namespace audio_utils::channels;
+    for (size_t i = 0; i < std::size(kSideFromChannelIdx); ++i) {
+        if ((mask & (1 << i)) != 0 && kSideFromChannelIdx[i] == AUDIO_GEOMETRY_SIDE_CENTER) {
+            return true;
+        }
+    }
+    return false;
+}
+
+/*
+ * Applies stereo volume to the audio data based on proper left right channel affinity
+ * (templated channel MASK parameter).
+ */
+template <int MIXTYPE, audio_channel_mask_t MASK,
         typename TO, typename TI, typename TV,
         typename F>
-void stereoVolumeHelper(TO*& out, const TI*& in, const TV *vol, F f) {
-    static_assert(NCHAN > 0 && NCHAN <= FCC_LIMIT);
-    static_assert(MIXTYPE == MIXTYPE_MULTI_STEREOVOL
-            || MIXTYPE == MIXTYPE_MULTI_SAVEONLY_STEREOVOL
-            || MIXTYPE == MIXTYPE_STEREOEXPAND
-            || MIXTYPE == MIXTYPE_MONOEXPAND);
+void stereoVolumeHelperWithChannelMask(TO*& out, const TI*& in, const TV *vol, F f) {
     auto proc = [](auto& a, const auto& b) {
         if constexpr (MIXTYPE == MIXTYPE_MULTI_STEREOVOL
                 || MIXTYPE == MIXTYPE_STEREOEXPAND
@@ -250,59 +263,113 @@
     auto inp = [&in]() -> const TI& {
         if constexpr (MIXTYPE == MIXTYPE_STEREOEXPAND
                 || MIXTYPE == MIXTYPE_MONOEXPAND) {
-            return *in;
+            return *in; // note STEREOEXPAND assumes replicated L/R channels (see doc below).
         } else {
             return *in++;
         }
     };
 
-    // HALs should only expose the canonical channel masks.
-    proc(*out++, f(inp(), vol[0])); // front left
-    if constexpr (NCHAN == 1) return;
-    proc(*out++, f(inp(), vol[1])); // front right
-    if constexpr (NCHAN == 2)  return;
-    if constexpr (NCHAN == 4) {
-        proc(*out++, f(inp(), vol[0])); // back left
-        proc(*out++, f(inp(), vol[1])); // back right
-        return;
-    }
-
-    // TODO: Precompute center volume if not ramping.
     std::decay_t<TV> center;
-    if constexpr (std::is_floating_point_v<TV>) {
-        center = (vol[0] + vol[1]) * 0.5;       // do not use divide
-    } else {
-        center = (vol[0] >> 1) + (vol[1] >> 1); // rounds to 0.
-    }
-    proc(*out++, f(inp(), center)); // center (or 2.1 LFE)
-    if constexpr (NCHAN == 3) return;
-    if constexpr (NCHAN == 5) {
-        proc(*out++, f(inp(), vol[0]));  // back left
-        proc(*out++, f(inp(), vol[1]));  // back right
-        return;
-    }
-
-    proc(*out++, f(inp(), center)); // lfe
-    proc(*out++, f(inp(), vol[0])); // back left
-    proc(*out++, f(inp(), vol[1])); // back right
-    if constexpr (NCHAN == 6) return;
-    if constexpr (NCHAN == 7) {
-        proc(*out++, f(inp(), center)); // back center
-        return;
-    }
-    // NCHAN == 8
-    proc(*out++, f(inp(), vol[0])); // side left
-    proc(*out++, f(inp(), vol[1])); // side right
-    if constexpr (NCHAN > FCC_8) {
-        // Mutes to zero extended surround channels.
-        // 7.1.4 has the correct behavior.
-        // 22.2 has the behavior that FLC and FRC will be mixed instead
-        // of SL and SR and LFE will be center, not left.
-        for (int i = 8; i < NCHAN; ++i) {
-            // TODO: Consider using android::audio_utils::channels::kSideFromChannelIdx
-            proc(*out++, f(inp(), 0.f));
+    constexpr bool USES_CENTER_CHANNEL = usesCenterChannel(MASK);
+    if constexpr (USES_CENTER_CHANNEL) {
+        if constexpr (std::is_floating_point_v<TV>) {
+            center = (vol[0] + vol[1]) * 0.5;       // do not use divide
+        } else {
+            center = (vol[0] >> 1) + (vol[1] >> 1); // rounds to 0.
         }
     }
+
+    using namespace audio_utils::channels;
+
+    // if LFE and LFE2 are both present, they take left and right volume respectively.
+    constexpr unsigned LFE_LFE2 = \
+             AUDIO_CHANNEL_OUT_LOW_FREQUENCY | AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2;
+    constexpr bool has_LFE_LFE2 = (MASK & LFE_LFE2) == LFE_LFE2;
+
+#pragma push_macro("DO_CHANNEL_POSITION")
+#undef DO_CHANNEL_POSITION
+#define DO_CHANNEL_POSITION(BIT_INDEX) \
+    if constexpr ((MASK & (1 << BIT_INDEX)) != 0) { \
+        constexpr auto side = kSideFromChannelIdx[BIT_INDEX]; \
+        if constexpr (side == AUDIO_GEOMETRY_SIDE_LEFT || \
+               has_LFE_LFE2 && (1 << BIT_INDEX) == AUDIO_CHANNEL_OUT_LOW_FREQUENCY) { \
+            proc(*out++, f(inp(), vol[0])); \
+        } else if constexpr (side == AUDIO_GEOMETRY_SIDE_RIGHT || \
+               has_LFE_LFE2 && (1 << BIT_INDEX) == AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2) { \
+            proc(*out++, f(inp(), vol[1])); \
+        } else /* constexpr */ { \
+            proc(*out++, f(inp(), center)); \
+        } \
+    }
+
+    DO_CHANNEL_POSITION(0);
+    DO_CHANNEL_POSITION(1);
+    DO_CHANNEL_POSITION(2);
+    DO_CHANNEL_POSITION(3);
+    DO_CHANNEL_POSITION(4);
+    DO_CHANNEL_POSITION(5);
+    DO_CHANNEL_POSITION(6);
+    DO_CHANNEL_POSITION(7);
+
+    DO_CHANNEL_POSITION(8);
+    DO_CHANNEL_POSITION(9);
+    DO_CHANNEL_POSITION(10);
+    DO_CHANNEL_POSITION(11);
+    DO_CHANNEL_POSITION(12);
+    DO_CHANNEL_POSITION(13);
+    DO_CHANNEL_POSITION(14);
+    DO_CHANNEL_POSITION(15);
+
+    DO_CHANNEL_POSITION(16);
+    DO_CHANNEL_POSITION(17);
+    DO_CHANNEL_POSITION(18);
+    DO_CHANNEL_POSITION(19);
+    DO_CHANNEL_POSITION(20);
+    DO_CHANNEL_POSITION(21);
+    DO_CHANNEL_POSITION(22);
+    DO_CHANNEL_POSITION(23);
+    DO_CHANNEL_POSITION(24);
+    DO_CHANNEL_POSITION(25);
+    static_assert(FCC_LIMIT <= FCC_26); // Note: this may need to change.
+#pragma pop_macro("DO_CHANNEL_POSITION")
+}
+
+// These are the channel position masks we expect from the HAL.
+// See audio_channel_out_mask_from_count() but this is constexpr
+constexpr inline audio_channel_mask_t canonicalChannelMaskFromCount(size_t channelCount) {
+    constexpr audio_channel_mask_t canonical[] = {
+        [0] = AUDIO_CHANNEL_NONE,
+        [1] = AUDIO_CHANNEL_OUT_MONO,
+        [2] = AUDIO_CHANNEL_OUT_STEREO,
+        [3] = AUDIO_CHANNEL_OUT_2POINT1,
+        [4] = AUDIO_CHANNEL_OUT_QUAD,
+        [5] = AUDIO_CHANNEL_OUT_PENTA,
+        [6] = AUDIO_CHANNEL_OUT_5POINT1,
+        [7] = AUDIO_CHANNEL_OUT_6POINT1,
+        [8] = AUDIO_CHANNEL_OUT_7POINT1,
+        [12] = AUDIO_CHANNEL_OUT_7POINT1POINT4,
+        [14] = AUDIO_CHANNEL_OUT_9POINT1POINT4,
+        [16] = AUDIO_CHANNEL_OUT_9POINT1POINT6,
+        [24] = AUDIO_CHANNEL_OUT_22POINT2,
+    };
+    return channelCount < std::size(canonical) ? canonical[channelCount] : AUDIO_CHANNEL_NONE;
+}
+
+template <int MIXTYPE, int NCHAN,
+        typename TO, typename TI, typename TV,
+        typename F>
+void stereoVolumeHelper(TO*& out, const TI*& in, const TV *vol, F f) {
+    static_assert(NCHAN > 0 && NCHAN <= FCC_LIMIT);
+    static_assert(MIXTYPE == MIXTYPE_MULTI_STEREOVOL
+            || MIXTYPE == MIXTYPE_MULTI_SAVEONLY_STEREOVOL
+            || MIXTYPE == MIXTYPE_STEREOEXPAND
+            || MIXTYPE == MIXTYPE_MONOEXPAND);
+    constexpr audio_channel_mask_t MASK{canonicalChannelMaskFromCount(NCHAN)};
+    if constexpr (MASK == AUDIO_CHANNEL_NONE) {
+        ALOGE("%s: Invalid position count %d", __func__, NCHAN);
+        return; // not a valid system mask, ignore.
+    }
+    stereoVolumeHelperWithChannelMask<MIXTYPE, MASK, TO, TI, TV, F>(out, in, vol, f);
 }
 
 /*
diff --git a/media/libaudioprocessing/include/media/AudioMixer.h b/media/libaudioprocessing/include/media/AudioMixer.h
index 70eafe3..5a9fa07 100644
--- a/media/libaudioprocessing/include/media/AudioMixer.h
+++ b/media/libaudioprocessing/include/media/AudioMixer.h
@@ -50,6 +50,7 @@
         // for haptic
         HAPTIC_ENABLED  = 0x4007, // Set haptic data from this track should be played or not.
         HAPTIC_INTENSITY = 0x4008, // Set the intensity to play haptic data.
+        HAPTIC_MAX_AMPLITUDE = 0x4009, // Set the max amplitude allowed for haptic data.
         // for target TIMESTRETCH
         PLAYBACK_RATE   = 0x4300, // Configure timestretch on this track name;
                                   // parameter 'value' is a pointer to the new playback rate.
@@ -145,6 +146,7 @@
         // Haptic
         bool                 mHapticPlaybackEnabled;
         os::HapticScale      mHapticIntensity;
+        float                mHapticMaxAmplitude;
         audio_channel_mask_t mHapticChannelMask;
         uint32_t             mHapticChannelCount;
         audio_channel_mask_t mMixerHapticChannelMask;
diff --git a/media/libaudioprocessing/tests/Android.bp b/media/libaudioprocessing/tests/Android.bp
index 3856817..ad402db 100644
--- a/media/libaudioprocessing/tests/Android.bp
+++ b/media/libaudioprocessing/tests/Android.bp
@@ -76,6 +76,7 @@
 //
 cc_binary {
     name: "mixerops_objdump",
+    header_libs: ["libaudioutils_headers"],
     srcs: ["mixerops_objdump.cpp"],
 }
 
@@ -84,6 +85,16 @@
 //
 cc_benchmark {
     name: "mixerops_benchmark",
+    header_libs: ["libaudioutils_headers"],
     srcs: ["mixerops_benchmark.cpp"],
     static_libs: ["libgoogle-benchmark"],
 }
+
+//
+// mixerops unit test
+//
+cc_test {
+    name: "mixerops_tests",
+    defaults: ["libaudioprocessing_test_defaults"],
+    srcs: ["mixerops_tests.cpp"],
+}
diff --git a/media/libaudioprocessing/tests/mixerops_benchmark.cpp b/media/libaudioprocessing/tests/mixerops_benchmark.cpp
index 7a4c5c7..f866b1a 100644
--- a/media/libaudioprocessing/tests/mixerops_benchmark.cpp
+++ b/media/libaudioprocessing/tests/mixerops_benchmark.cpp
@@ -16,11 +16,9 @@
 
 #include <inttypes.h>
 #include <type_traits>
-#include "../../../../system/media/audio_utils/include/audio_utils/primitives.h"
 #define LOG_ALWAYS_FATAL(...)
 
 #include <../AudioMixerOps.h>
-
 #include <benchmark/benchmark.h>
 
 using namespace android;
diff --git a/media/libaudioprocessing/tests/mixerops_tests.cpp b/media/libaudioprocessing/tests/mixerops_tests.cpp
new file mode 100644
index 0000000..2500ba9
--- /dev/null
+++ b/media/libaudioprocessing/tests/mixerops_tests.cpp
@@ -0,0 +1,175 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+//#define LOG_NDEBUG 0
+#define LOG_TAG "mixerop_tests"
+#include <log/log.h>
+
+#include <inttypes.h>
+#include <type_traits>
+
+#include <../AudioMixerOps.h>
+#include <gtest/gtest.h>
+
+using namespace android;
+
+// Note: gtest templated tests require typenames, not integers.
+template <int MIXTYPE, int NCHAN>
+class MixerOpsBasicTest {
+public:
+    static void testStereoVolume() {
+        using namespace android::audio_utils::channels;
+
+        constexpr size_t FRAME_COUNT = 1000;
+        constexpr size_t SAMPLE_COUNT = FRAME_COUNT * NCHAN;
+
+        const float in[SAMPLE_COUNT] = {[0 ... (SAMPLE_COUNT - 1)] = 1.f};
+
+        AUDIO_GEOMETRY_SIDE sides[NCHAN];
+        size_t i = 0;
+        unsigned channel = canonicalChannelMaskFromCount(NCHAN);
+        constexpr unsigned LFE_LFE2 =
+                AUDIO_CHANNEL_OUT_LOW_FREQUENCY | AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2;
+        bool has_LFE_LFE2 = (channel & LFE_LFE2) == LFE_LFE2;
+        while (channel != 0) {
+            const int index = __builtin_ctz(channel);
+            if (has_LFE_LFE2 && (1 << index) == AUDIO_CHANNEL_OUT_LOW_FREQUENCY) {
+                sides[i++] = AUDIO_GEOMETRY_SIDE_LEFT; // special case
+            } else if (has_LFE_LFE2 && (1 << index) == AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2) {
+                sides[i++] = AUDIO_GEOMETRY_SIDE_RIGHT; // special case
+            } else {
+                sides[i++] = sideFromChannelIdx(index);
+            }
+            channel &= ~(1 << index);
+        }
+
+        float vola[2] = {1.f, 0.f}; // left volume at max.
+        float out[SAMPLE_COUNT]{};
+        float aux[FRAME_COUNT]{};
+        float volaux = 0.5;
+        {
+            volumeMulti<MIXTYPE, NCHAN>(out, FRAME_COUNT, in, aux, vola, volaux);
+            const float *outp = out;
+            const float *auxp = aux;
+            const float left = vola[0];
+            const float center = (vola[0] + vola[1]) * 0.5;
+            const float right = vola[1];
+            for (size_t i = 0; i < FRAME_COUNT; ++i) {
+                for (size_t j = 0; j < NCHAN; ++j) {
+                    const float audio = *outp++;
+                    if (sides[j] == AUDIO_GEOMETRY_SIDE_LEFT) {
+                        EXPECT_EQ(left, audio);
+                    } else if (sides[j] == AUDIO_GEOMETRY_SIDE_CENTER) {
+                        EXPECT_EQ(center, audio);
+                    } else {
+                        EXPECT_EQ(right, audio);
+                    }
+                }
+                EXPECT_EQ(volaux, *auxp++);  // works if all channels contain 1.f
+            }
+        }
+        float volb[2] = {0.f, 0.5f}; // right volume at half max.
+        {
+            // this accumulates into out, aux.
+            // float out[SAMPLE_COUNT]{};
+            // float aux[FRAME_COUNT]{};
+            volumeMulti<MIXTYPE, NCHAN>(out, FRAME_COUNT, in, aux, volb, volaux);
+            const float *outp = out;
+            const float *auxp = aux;
+            const float left = vola[0] + volb[0];
+            const float center = (vola[0] + vola[1] + volb[0] + volb[1]) * 0.5;
+            const float right = vola[1] + volb[1];
+            for (size_t i = 0; i < FRAME_COUNT; ++i) {
+                for (size_t j = 0; j < NCHAN; ++j) {
+                    const float audio = *outp++;
+                    if (sides[j] == AUDIO_GEOMETRY_SIDE_LEFT) {
+                        EXPECT_EQ(left, audio);
+                    } else if (sides[j] == AUDIO_GEOMETRY_SIDE_CENTER) {
+                        EXPECT_EQ(center, audio);
+                    } else {
+                        EXPECT_EQ(right, audio);
+                    }
+                }
+                // aux is accumulated so 2x the amplitude
+                EXPECT_EQ(volaux * 2.f, *auxp++);  // works if all channels contain 1.f
+            }
+        }
+
+        { // test aux as derived from out.
+            // AUX channel is the weighted sum of all of the output channels prior to volume
+            // adjustment.  We must set L and R to the same volume to allow computation
+            // of AUX from the output values.
+            const float volmono = 0.25f;
+            const float vollr[2] = {volmono, volmono}; // all the same.
+            float out[SAMPLE_COUNT]{};
+            float aux[FRAME_COUNT]{};
+            volumeMulti<MIXTYPE, NCHAN>(out, FRAME_COUNT, in, aux, vollr, volaux);
+            const float *outp = out;
+            const float *auxp = aux;
+            for (size_t i = 0; i < FRAME_COUNT; ++i) {
+                float accum = 0.f;
+                for (size_t j = 0; j < NCHAN; ++j) {
+                    accum += *outp++;
+                }
+                EXPECT_EQ(accum / NCHAN * volaux / volmono, *auxp++);
+            }
+        }
+    }
+};
+
+TEST(mixerops, stereovolume_1) { // Note: mono not used for output sinks yet.
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 1>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_2) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 2>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_3) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 3>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_4) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 4>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_5) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 5>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_6) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 6>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_7) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 7>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_8) {
+    MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 8>::testStereoVolume();
+}
+TEST(mixerops, stereovolume_12) {
+    if constexpr (FCC_LIMIT >= 12) { // NOTE: FCC_LIMIT is an enum, so can't #if
+        MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 12>::testStereoVolume();
+    }
+}
+TEST(mixerops, stereovolume_24) {
+    if constexpr (FCC_LIMIT >= 24) {
+        MixerOpsBasicTest<MIXTYPE_MULTI_STEREOVOL, 24>::testStereoVolume();
+    }
+}
+TEST(mixerops, channel_equivalence) {
+    // we must match the constexpr function with the system determined channel mask from count.
+    for (size_t i = 0; i < FCC_LIMIT; ++i) {
+        const audio_channel_mask_t actual = canonicalChannelMaskFromCount(i);
+        const audio_channel_mask_t system = audio_channel_out_mask_from_count(i);
+        if (system == AUDIO_CHANNEL_INVALID) continue;
+        EXPECT_EQ(system, actual);
+    }
+}
diff --git a/media/libeffects/downmix/EffectDownmix.cpp b/media/libeffects/downmix/EffectDownmix.cpp
index f500bc3..90bb410 100644
--- a/media/libeffects/downmix/EffectDownmix.cpp
+++ b/media/libeffects/downmix/EffectDownmix.cpp
@@ -19,7 +19,7 @@
 #include <log/log.h>
 
 #include "EffectDownmix.h"
-#include <math.h>
+#include <audio_utils/ChannelMix.h>
 
 // Do not submit with DOWNMIX_TEST_CHANNEL_INDEX defined, strictly for testing
 //#define DOWNMIX_TEST_CHANNEL_INDEX 0
@@ -35,12 +35,13 @@
 } downmix_state_t;
 
 /* parameters for each downmixer */
-typedef struct {
+struct downmix_object_t {
     downmix_state_t state;
     downmix_type_t type;
     bool apply_volume_correction;
     uint8_t input_channel_count;
-} downmix_object_t;
+    android::audio_utils::channels::ChannelMix channelMix;
+};
 
 typedef struct downmix_module_s {
     const struct effect_interface_s *itfe;
@@ -77,11 +78,6 @@
         downmix_object_t *pDownmixer, int32_t param, uint32_t size, void *pValue);
 static int Downmix_getParameter(
         downmix_object_t *pDownmixer, int32_t param, uint32_t *pSize, void *pValue);
-static void Downmix_foldFromQuad(float *pSrc, float *pDst, size_t numFrames, bool accumulate);
-static void Downmix_foldFrom5Point1(float *pSrc, float *pDst, size_t numFrames, bool accumulate);
-static void Downmix_foldFrom7Point1(float *pSrc, float *pDst, size_t numFrames, bool accumulate);
-static bool Downmix_foldGeneric(
-        uint32_t mask, float *pSrc, float *pDst, size_t numFrames, bool accumulate);
 
 // effect_handle_t interface implementation for downmix effect
 const struct effect_interface_s gDownmixInterface = {
@@ -315,7 +311,8 @@
         audio_buffer_t *inBuffer, audio_buffer_t *outBuffer) {
 
     downmix_object_t *pDownmixer;
-    float *pSrc, *pDst;
+    const float *pSrc;
+    float *pDst;
     downmix_module_t *pDwmModule = (downmix_module_t *)self;
 
     if (pDwmModule == NULL) {
@@ -344,7 +341,8 @@
 
     const bool accumulate =
             (pDwmModule->config.outputCfg.accessMode == EFFECT_BUFFER_ACCESS_ACCUMULATE);
-    const uint32_t downmixInputChannelMask = pDwmModule->config.inputCfg.channels;
+    const audio_channel_mask_t downmixInputChannelMask =
+            (audio_channel_mask_t)pDwmModule->config.inputCfg.channels;
 
     switch(pDownmixer->type) {
 
@@ -368,38 +366,13 @@
           }
           break;
 
-      case DOWNMIX_TYPE_FOLD:
-#ifdef DOWNMIX_ALWAYS_USE_GENERIC_DOWNMIXER
-          // bypass the optimized downmix routines for the common formats
-          if (!Downmix_foldGeneric(
-                  downmixInputChannelMask, pSrc, pDst, numFrames, accumulate)) {
-              ALOGE("Multichannel configuration %#x is not supported",
-                    downmixInputChannelMask);
-              return -EINVAL;
-          }
-          break;
-#endif
-        // optimize for the common formats
-        switch (downmixInputChannelMask) {
-        case AUDIO_CHANNEL_OUT_QUAD_BACK:
-        case AUDIO_CHANNEL_OUT_QUAD_SIDE:
-            Downmix_foldFromQuad(pSrc, pDst, numFrames, accumulate);
-            break;
-        case AUDIO_CHANNEL_OUT_5POINT1_BACK:
-        case AUDIO_CHANNEL_OUT_5POINT1_SIDE:
-            Downmix_foldFrom5Point1(pSrc, pDst, numFrames, accumulate);
-            break;
-        case AUDIO_CHANNEL_OUT_7POINT1:
-            Downmix_foldFrom7Point1(pSrc, pDst, numFrames, accumulate);
-            break;
-        default:
-            if (!Downmix_foldGeneric(
-                    downmixInputChannelMask, pSrc, pDst, numFrames, accumulate)) {
+      case DOWNMIX_TYPE_FOLD: {
+            if (!pDownmixer->channelMix.process(
+                    pSrc, pDst, numFrames, accumulate, downmixInputChannelMask)) {
                 ALOGE("Multichannel configuration %#x is not supported",
                       downmixInputChannelMask);
                 return -EINVAL;
             }
-            break;
         }
         break;
 
@@ -780,7 +753,6 @@
     return 0;
 } /* end Downmix_setParameter */
 
-
 /*----------------------------------------------------------------------------
  * Downmix_getParameter()
  *----------------------------------------------------------------------------
@@ -829,299 +801,3 @@
     return 0;
 } /* end Downmix_getParameter */
 
-
-/*----------------------------------------------------------------------------
- * Downmix_foldFromQuad()
- *----------------------------------------------------------------------------
- * Purpose:
- * downmix a quad signal to stereo
- *
- * Inputs:
- *  pSrc       quad audio samples to downmix
- *  numFrames  the number of quad frames to downmix
- *  accumulate whether to mix (when true) the result of the downmix with the contents of pDst,
- *               or overwrite pDst (when false)
- *
- * Outputs:
- *  pDst       downmixed stereo audio samples
- *
- *----------------------------------------------------------------------------
- */
-void Downmix_foldFromQuad(float *pSrc, float *pDst, size_t numFrames, bool accumulate) {
-    // sample at index 0 is FL
-    // sample at index 1 is FR
-    // sample at index 2 is RL
-    // sample at index 3 is RR
-    if (accumulate) {
-        while (numFrames) {
-            // FL + RL
-            pDst[0] = clamp_float(pDst[0] + ((pSrc[0] + pSrc[2]) / 2.0f));
-            // FR + RR
-            pDst[1] = clamp_float(pDst[1] + ((pSrc[1] + pSrc[3]) / 2.0f));
-            pSrc += 4;
-            pDst += 2;
-            numFrames--;
-        }
-    } else { // same code as above but without adding and clamping pDst[i] to itself
-        while (numFrames) {
-            // FL + RL
-            pDst[0] = clamp_float((pSrc[0] + pSrc[2]) / 2.0f);
-            // FR + RR
-            pDst[1] = clamp_float((pSrc[1] + pSrc[3]) / 2.0f);
-            pSrc += 4;
-            pDst += 2;
-            numFrames--;
-        }
-    }
-}
-
-/*----------------------------------------------------------------------------
- * Downmix_foldFrom5Point1()
- *----------------------------------------------------------------------------
- * Purpose:
- * downmix a 5.1 signal to stereo
- *
- * Inputs:
- *  pSrc       5.1 audio samples to downmix
- *  numFrames  the number of 5.1 frames to downmix
- *  accumulate whether to mix (when true) the result of the downmix with the contents of pDst,
- *               or overwrite pDst (when false)
- *
- * Outputs:
- *  pDst       downmixed stereo audio samples
- *
- *----------------------------------------------------------------------------
- */
-void Downmix_foldFrom5Point1(float *pSrc, float *pDst, size_t numFrames, bool accumulate) {
-    float lt, rt, centerPlusLfeContrib; // samples in Q19.12 format
-    // sample at index 0 is FL
-    // sample at index 1 is FR
-    // sample at index 2 is FC
-    // sample at index 3 is LFE
-    // sample at index 4 is RL
-    // sample at index 5 is RR
-    // code is mostly duplicated between the two values of accumulate to avoid repeating the test
-    // for every sample
-    if (accumulate) {
-        while (numFrames) {
-            // centerPlusLfeContrib = FC(-3dB) + LFE(-3dB)
-            centerPlusLfeContrib = (pSrc[2] * MINUS_3_DB_IN_FLOAT)
-                    + (pSrc[3] * MINUS_3_DB_IN_FLOAT);
-            // FL + centerPlusLfeContrib + RL
-            lt = pSrc[0] + centerPlusLfeContrib + pSrc[4];
-            // FR + centerPlusLfeContrib + RR
-            rt = pSrc[1] + centerPlusLfeContrib + pSrc[5];
-            // accumulate in destination
-            pDst[0] = clamp_float(pDst[0] + (lt / 2.0f));
-            pDst[1] = clamp_float(pDst[1] + (rt / 2.0f));
-            pSrc += 6;
-            pDst += 2;
-            numFrames--;
-        }
-    } else { // same code as above but without adding and clamping pDst[i] to itself
-        while (numFrames) {
-            // centerPlusLfeContrib = FC(-3dB) + LFE(-3dB)
-            centerPlusLfeContrib = (pSrc[2] * MINUS_3_DB_IN_FLOAT)
-                    + (pSrc[3] * MINUS_3_DB_IN_FLOAT);
-            // FL + centerPlusLfeContrib + RL
-            lt = pSrc[0] + centerPlusLfeContrib + pSrc[4];
-            // FR + centerPlusLfeContrib + RR
-            rt = pSrc[1] + centerPlusLfeContrib + pSrc[5];
-            // store in destination
-            pDst[0] = clamp_float(lt / 2.0f); // differs from when accumulate is true above
-            pDst[1] = clamp_float(rt / 2.0f); // differs from when accumulate is true above
-            pSrc += 6;
-            pDst += 2;
-            numFrames--;
-        }
-    }
-}
-
-/*----------------------------------------------------------------------------
- * Downmix_foldFrom7Point1()
- *----------------------------------------------------------------------------
- * Purpose:
- * downmix a 7.1 signal to stereo
- *
- * Inputs:
- *  pSrc       7.1 audio samples to downmix
- *  numFrames  the number of 7.1 frames to downmix
- *  accumulate whether to mix (when true) the result of the downmix with the contents of pDst,
- *               or overwrite pDst (when false)
- *
- * Outputs:
- *  pDst       downmixed stereo audio samples
- *
- *----------------------------------------------------------------------------
- */
-void Downmix_foldFrom7Point1(float *pSrc, float *pDst, size_t numFrames, bool accumulate) {
-    float lt, rt, centerPlusLfeContrib; // samples in Q19.12 format
-    // sample at index 0 is FL
-    // sample at index 1 is FR
-    // sample at index 2 is FC
-    // sample at index 3 is LFE
-    // sample at index 4 is RL
-    // sample at index 5 is RR
-    // sample at index 6 is SL
-    // sample at index 7 is SR
-    // code is mostly duplicated between the two values of accumulate to avoid repeating the test
-    // for every sample
-    if (accumulate) {
-        while (numFrames) {
-            // centerPlusLfeContrib = FC(-3dB) + LFE(-3dB)
-            centerPlusLfeContrib = (pSrc[2] * MINUS_3_DB_IN_FLOAT)
-                    + (pSrc[3] * MINUS_3_DB_IN_FLOAT);
-            // FL + centerPlusLfeContrib + SL + RL
-            lt = pSrc[0] + centerPlusLfeContrib + pSrc[6] + pSrc[4];
-            // FR + centerPlusLfeContrib + SR + RR
-            rt = pSrc[1] + centerPlusLfeContrib + pSrc[7] + pSrc[5];
-            //accumulate in destination
-            pDst[0] = clamp_float(pDst[0] + (lt / 2.0f));
-            pDst[1] = clamp_float(pDst[1] + (rt / 2.0f));
-            pSrc += 8;
-            pDst += 2;
-            numFrames--;
-        }
-    } else { // same code as above but without adding and clamping pDst[i] to itself
-        while (numFrames) {
-            // centerPlusLfeContrib = FC(-3dB) + LFE(-3dB)
-            centerPlusLfeContrib = (pSrc[2] * MINUS_3_DB_IN_FLOAT)
-                    + (pSrc[3] * MINUS_3_DB_IN_FLOAT);
-            // FL + centerPlusLfeContrib + SL + RL
-            lt = pSrc[0] + centerPlusLfeContrib + pSrc[6] + pSrc[4];
-            // FR + centerPlusLfeContrib + SR + RR
-            rt = pSrc[1] + centerPlusLfeContrib + pSrc[7] + pSrc[5];
-            // store in destination
-            pDst[0] = clamp_float(lt / 2.0f); // differs from when accumulate is true above
-            pDst[1] = clamp_float(rt / 2.0f); // differs from when accumulate is true above
-            pSrc += 8;
-            pDst += 2;
-            numFrames--;
-        }
-    }
-}
-
-/*----------------------------------------------------------------------------
- * Downmix_foldGeneric()
- *----------------------------------------------------------------------------
- * Purpose:
- * downmix to stereo a multichannel signal of arbitrary channel position mask.
- *
- * Inputs:
- *  mask       the channel mask of pSrc
- *  pSrc       multichannel audio buffer to downmix
- *  numFrames  the number of multichannel frames to downmix
- *  accumulate whether to mix (when true) the result of the downmix with the contents of pDst,
- *               or overwrite pDst (when false)
- *
- * Outputs:
- *  pDst       downmixed stereo audio samples
- *
- * Returns: false if multichannel format is not supported
- *
- *----------------------------------------------------------------------------
- */
-bool Downmix_foldGeneric(
-        uint32_t mask, float *pSrc, float *pDst, size_t numFrames, bool accumulate) {
-
-    if (!Downmix_validChannelMask(mask)) {
-        return false;
-    }
-    const int numChan = audio_channel_count_from_out_mask(mask);
-
-    // compute at what index each channel is: samples will be in the following order:
-    //   FL  FR  FC    LFE   BL  BR  BC    SL  SR
-    //
-    //  (transfer matrix)
-    //   FL  FR  FC    LFE   BL  BR  BC    SL  SR
-    //   0.5     0.353 0.353 0.5     0.353 0.5
-    //       0.5 0.353 0.353     0.5 0.353     0.5
-
-    // derive the indices for the transfer matrix columns that have non-zero values.
-    int indexFL = -1;
-    int indexFR = -1;
-    int indexFC = -1;
-    int indexLFE = -1;
-    int indexBL = -1;
-    int indexBR = -1;
-    int indexBC = -1;
-    int indexSL = -1;
-    int indexSR = -1;
-    int index = 0;
-    for (unsigned tmp = mask;
-         (tmp & (AUDIO_CHANNEL_OUT_7POINT1 | AUDIO_CHANNEL_OUT_BACK_CENTER)) != 0;
-         ++index) {
-        const unsigned lowestBit = tmp & -(signed)tmp;
-        switch (lowestBit) {
-        case AUDIO_CHANNEL_OUT_FRONT_LEFT:
-            indexFL = index;
-            break;
-        case AUDIO_CHANNEL_OUT_FRONT_RIGHT:
-            indexFR = index;
-            break;
-        case AUDIO_CHANNEL_OUT_FRONT_CENTER:
-            indexFC = index;
-            break;
-        case AUDIO_CHANNEL_OUT_LOW_FREQUENCY:
-            indexLFE = index;
-            break;
-        case AUDIO_CHANNEL_OUT_BACK_LEFT:
-            indexBL = index;
-            break;
-        case AUDIO_CHANNEL_OUT_BACK_RIGHT:
-            indexBR = index;
-            break;
-        case AUDIO_CHANNEL_OUT_BACK_CENTER:
-            indexBC = index;
-            break;
-        case AUDIO_CHANNEL_OUT_SIDE_LEFT:
-            indexSL = index;
-            break;
-        case AUDIO_CHANNEL_OUT_SIDE_RIGHT:
-            indexSR = index;
-            break;
-        }
-        tmp ^= lowestBit;
-    }
-
-    // With good branch prediction, this should run reasonably fast.
-    // Also consider using a transfer matrix form.
-    while (numFrames) {
-        // compute contribution of FC, BC and LFE
-        float centersLfeContrib = 0;
-        if (indexFC >= 0) centersLfeContrib = pSrc[indexFC];
-        if (indexLFE >= 0) centersLfeContrib += pSrc[indexLFE];
-        if (indexBC >= 0) centersLfeContrib += pSrc[indexBC];
-        centersLfeContrib *= MINUS_3_DB_IN_FLOAT;
-
-        float ch[2];
-        ch[0] = centersLfeContrib;
-        ch[1] = centersLfeContrib;
-
-        // mix in left / right channels
-        if (indexFL >= 0) ch[0] += pSrc[indexFL];
-        if (indexFR >= 0) ch[1] += pSrc[indexFR];
-
-        if (indexSL >= 0) ch[0] += pSrc[indexSL];
-        if (indexSR >= 0) ch[1] += pSrc[indexSR]; // note pair checks enforce this if indexSL != 0
-
-        if (indexBL >= 0) ch[0] += pSrc[indexBL];
-        if (indexBR >= 0) ch[1] += pSrc[indexBR]; // note pair checks enforce this if indexBL != 0
-
-        // scale to prevent overflow.
-        ch[0] *= 0.5f;
-        ch[1] *= 0.5f;
-
-        if (accumulate) {
-            ch[0] += pDst[0];
-            ch[1] += pDst[1];
-        }
-
-        pDst[0] = clamp_float(ch[0]);
-        pDst[1] = clamp_float(ch[1]);
-        pSrc += numChan;
-        pDst += 2;
-        numFrames--;
-    }
-    return true;
-}
diff --git a/media/libeffects/downmix/benchmark/downmix_benchmark.cpp b/media/libeffects/downmix/benchmark/downmix_benchmark.cpp
index ee169c2..d9d40ed 100644
--- a/media/libeffects/downmix/benchmark/downmix_benchmark.cpp
+++ b/media/libeffects/downmix/benchmark/downmix_benchmark.cpp
@@ -35,16 +35,14 @@
     AUDIO_CHANNEL_OUT_STEREO,
     AUDIO_CHANNEL_OUT_2POINT1,
     AUDIO_CHANNEL_OUT_2POINT0POINT2,
-    AUDIO_CHANNEL_OUT_QUAD,
-    AUDIO_CHANNEL_OUT_QUAD_BACK,
+    AUDIO_CHANNEL_OUT_QUAD, // AUDIO_CHANNEL_OUT_QUAD_BACK
     AUDIO_CHANNEL_OUT_QUAD_SIDE,
     AUDIO_CHANNEL_OUT_SURROUND,
     AUDIO_CHANNEL_OUT_2POINT1POINT2,
     AUDIO_CHANNEL_OUT_3POINT0POINT2,
     AUDIO_CHANNEL_OUT_PENTA,
     AUDIO_CHANNEL_OUT_3POINT1POINT2,
-    AUDIO_CHANNEL_OUT_5POINT1,
-    AUDIO_CHANNEL_OUT_5POINT1_BACK,
+    AUDIO_CHANNEL_OUT_5POINT1, // AUDIO_CHANNEL_OUT_5POINT1_BACK
     AUDIO_CHANNEL_OUT_5POINT1_SIDE,
     AUDIO_CHANNEL_OUT_6POINT1,
     AUDIO_CHANNEL_OUT_5POINT1POINT2,
@@ -62,58 +60,34 @@
 static constexpr size_t kFrameCount = 1000;
 
 /*
-Pixel 3XL
-downmix_benchmark:
-  #BM_Downmix/0     4723 ns    4708 ns       148694
-  #BM_Downmix/1     4717 ns    4702 ns       148873
-  #BM_Downmix/2     4803 ns    4788 ns       145893
-  #BM_Downmix/3     5056 ns    5041 ns       139110
-  #BM_Downmix/4     4710 ns    4696 ns       149625
-  #BM_Downmix/5     1514 ns    1509 ns       463694
-  #BM_Downmix/6     1513 ns    1509 ns       463451
-  #BM_Downmix/7     1516 ns    1511 ns       463899
-  #BM_Downmix/8     4445 ns    4431 ns       157831
-  #BM_Downmix/9     5081 ns    5065 ns       138412
-  #BM_Downmix/10    4354 ns    4341 ns       161247
-  #BM_Downmix/11    4411 ns    4397 ns       158893
-  #BM_Downmix/12    4434 ns    4420 ns       157992
-  #BM_Downmix/13    4845 ns    4830 ns       144873
-  #BM_Downmix/14    4851 ns    4835 ns       144954
-  #BM_Downmix/15    4884 ns    4870 ns       144233
-  #BM_Downmix/16    5832 ns    5813 ns       120565
-  #BM_Downmix/17    5241 ns    5224 ns       133927
-  #BM_Downmix/18    5044 ns    5028 ns       139131
-  #BM_Downmix/19    5244 ns    5227 ns       132315
-  #BM_Downmix/20    5943 ns    5923 ns       117759
-  #BM_Downmix/21    5990 ns    5971 ns       117263
-  #BM_Downmix/22    4468 ns    4454 ns       156689
-  #BM_Downmix/23    7306 ns    7286 ns        95911
---
-downmix_benchmark: (generic fold)
-  #BM_Downmix/0     4722 ns    4707 ns       149847
-  #BM_Downmix/1     4714 ns    4698 ns       148748
-  #BM_Downmix/2     4794 ns    4779 ns       145661
-  #BM_Downmix/3     5053 ns    5035 ns       139172
-  #BM_Downmix/4     4695 ns    4678 ns       149762
-  #BM_Downmix/5     4381 ns    4368 ns       159675
-  #BM_Downmix/6     4387 ns    4373 ns       160267
-  #BM_Downmix/7     4732 ns    4717 ns       148514
-  #BM_Downmix/8     4430 ns    4415 ns       158133
-  #BM_Downmix/9     5101 ns    5084 ns       138353
-  #BM_Downmix/10    4356 ns    4343 ns       160821
-  #BM_Downmix/11    4397 ns    4383 ns       159995
-  #BM_Downmix/12    4438 ns    4424 ns       158117
-  #BM_Downmix/13    5243 ns    5226 ns       133863
-  #BM_Downmix/14    5259 ns    5242 ns       131855
-  #BM_Downmix/15    5245 ns    5228 ns       133686
-  #BM_Downmix/16    5829 ns    5809 ns       120543
-  #BM_Downmix/17    5245 ns    5228 ns       133533
-  #BM_Downmix/18    5935 ns    5916 ns       118282
-  #BM_Downmix/19    5263 ns    5245 ns       133657
-  #BM_Downmix/20    5998 ns    5978 ns       114693
-  #BM_Downmix/21    5989 ns    5969 ns       117450
-  #BM_Downmix/22    4442 ns    4431 ns       157913
-  #BM_Downmix/23    7309 ns    7290 ns        95797
+Pixel 4XL
+$ adb shell /data/benchmarktest/downmix_benchmark/vendor/downmix_benchmark
+
+--------------------------------------------------------
+Benchmark              Time             CPU   Iterations
+--------------------------------------------------------
+BM_Downmix/0        3638 ns         3624 ns       197517 AUDIO_CHANNEL_OUT_MONO
+BM_Downmix/1        4040 ns         4024 ns       178766
+BM_Downmix/2        4759 ns         4740 ns       134741 AUDIO_CHANNEL_OUT_STEREO
+BM_Downmix/3        6042 ns         6017 ns       129546 AUDIO_CHANNEL_OUT_2POINT1
+BM_Downmix/4        6897 ns         6868 ns        96316 AUDIO_CHANNEL_OUT_2POINT0POINT2
+BM_Downmix/5        2117 ns         2109 ns       331705 AUDIO_CHANNEL_OUT_QUAD
+BM_Downmix/6        2097 ns         2088 ns       335421 AUDIO_CHANNEL_OUT_QUAD_SIDE
+BM_Downmix/7        7291 ns         7263 ns        96256 AUDIO_CHANNEL_OUT_SURROUND
+BM_Downmix/8        8246 ns         8206 ns        84318 AUDIO_CHANNEL_OUT_2POINT1POINT2
+BM_Downmix/9        8341 ns         8303 ns        84298 AUDIO_CHANNEL_OUT_3POINT0POINT2
+BM_Downmix/10       7549 ns         7517 ns        84293 AUDIO_CHANNEL_OUT_PENTA
+BM_Downmix/11       9395 ns         9354 ns        75209 AUDIO_CHANNEL_OUT_3POINT1POINT2
+BM_Downmix/12       3267 ns         3253 ns       215596 AUDIO_CHANNEL_OUT_5POINT1
+BM_Downmix/13       3178 ns         3163 ns       220132 AUDIO_CHANNEL_OUT_5POINT1_SIDE
+BM_Downmix/14      10245 ns        10199 ns        67486 AUDIO_CHANNEL_OUT_6POINT1
+BM_Downmix/15      10975 ns        10929 ns        61359 AUDIO_CHANNEL_OUT_5POINT1POINT2
+BM_Downmix/16       3796 ns         3780 ns       184728 AUDIO_CHANNEL_OUT_7POINT1
+BM_Downmix/17      13562 ns        13503 ns        51823 AUDIO_CHANNEL_OUT_5POINT1POINT4
+BM_Downmix/18      13573 ns        13516 ns        51800 AUDIO_CHANNEL_OUT_7POINT1POINT2
+BM_Downmix/19      15502 ns        15435 ns        47147 AUDIO_CHANNEL_OUT_7POINT1POINT4
+BM_Downmix/20      16693 ns        16624 ns        42109 AUDIO_CHANNEL_OUT_13POINT_360RA
+BM_Downmix/21      28267 ns        28116 ns        24982 AUDIO_CHANNEL_OUT_22POINT2
 */
 
 static void BM_Downmix(benchmark::State& state) {
@@ -125,7 +99,7 @@
     std::minstd_rand gen(channelMask);
     std::uniform_real_distribution<> dis(-1.0f, 1.0f);
     std::vector<float> input(kFrameCount * channelCount);
-    std::vector<float> output(kFrameCount * 2);
+    std::vector<float> output(kFrameCount * FCC_2);
     for (auto& in : input) {
         in = dis(gen);
     }
@@ -187,7 +161,8 @@
         benchmark::ClobberMemory();
     }
 
-    state.SetComplexityN(state.range(0));
+    state.SetComplexityN(channelCount);
+    state.SetLabel(audio_channel_out_mask_to_string(channelMask));
 
     if (int status = AUDIO_EFFECT_LIBRARY_INFO_SYM.release_effect(effectHandle); status != 0) {
         ALOGE("release_effect returned an error = %d\n", status);
diff --git a/media/libeffects/downmix/tests/downmix_tests.cpp b/media/libeffects/downmix/tests/downmix_tests.cpp
index d4b7a3a..26c7c31 100644
--- a/media/libeffects/downmix/tests/downmix_tests.cpp
+++ b/media/libeffects/downmix/tests/downmix_tests.cpp
@@ -33,16 +33,14 @@
     AUDIO_CHANNEL_OUT_STEREO,
     AUDIO_CHANNEL_OUT_2POINT1,
     AUDIO_CHANNEL_OUT_2POINT0POINT2,
-    AUDIO_CHANNEL_OUT_QUAD,
-    AUDIO_CHANNEL_OUT_QUAD_BACK,
+    AUDIO_CHANNEL_OUT_QUAD, // AUDIO_CHANNEL_OUT_QUAD_BACK
     AUDIO_CHANNEL_OUT_QUAD_SIDE,
     AUDIO_CHANNEL_OUT_SURROUND,
     AUDIO_CHANNEL_OUT_2POINT1POINT2,
     AUDIO_CHANNEL_OUT_3POINT0POINT2,
     AUDIO_CHANNEL_OUT_PENTA,
     AUDIO_CHANNEL_OUT_3POINT1POINT2,
-    AUDIO_CHANNEL_OUT_5POINT1,
-    AUDIO_CHANNEL_OUT_5POINT1_BACK,
+    AUDIO_CHANNEL_OUT_5POINT1, // AUDIO_CHANNEL_OUT_5POINT1_BACK
     AUDIO_CHANNEL_OUT_5POINT1_SIDE,
     AUDIO_CHANNEL_OUT_6POINT1,
     AUDIO_CHANNEL_OUT_5POINT1POINT2,
@@ -54,8 +52,64 @@
     AUDIO_CHANNEL_OUT_22POINT2,
 };
 
-static constexpr audio_channel_mask_t kConsideredChannels =
-    (audio_channel_mask_t)(AUDIO_CHANNEL_OUT_7POINT1 | AUDIO_CHANNEL_OUT_BACK_CENTER);
+constexpr float COEF_25 = 0.2508909536f;
+constexpr float COEF_35 = 0.3543928915f;
+constexpr float COEF_36 = 0.3552343859f;
+constexpr float COEF_61 = 0.6057043428f;
+
+constexpr inline float kScaleFromChannelIdxLeft[] = {
+    1.f,       // AUDIO_CHANNEL_OUT_FRONT_LEFT            = 0x1u,
+    0.f,       // AUDIO_CHANNEL_OUT_FRONT_RIGHT           = 0x2u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_FRONT_CENTER          = 0x4u,
+    0.5f,      // AUDIO_CHANNEL_OUT_LOW_FREQUENCY         = 0x8u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_BACK_LEFT             = 0x10u,
+    0.f,       // AUDIO_CHANNEL_OUT_BACK_RIGHT            = 0x20u,
+    COEF_61,   // AUDIO_CHANNEL_OUT_FRONT_LEFT_OF_CENTER  = 0x40u,
+    COEF_25,   // AUDIO_CHANNEL_OUT_FRONT_RIGHT_OF_CENTER = 0x80u,
+    0.5f,      // AUDIO_CHANNEL_OUT_BACK_CENTER           = 0x100u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_SIDE_LEFT             = 0x200u,
+    0.f,       // AUDIO_CHANNEL_OUT_SIDE_RIGHT            = 0x400u,
+    COEF_36,   // AUDIO_CHANNEL_OUT_TOP_CENTER            = 0x800u,
+    1.f,       // AUDIO_CHANNEL_OUT_TOP_FRONT_LEFT        = 0x1000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_TOP_FRONT_CENTER      = 0x2000u,
+    0.f,       // AUDIO_CHANNEL_OUT_TOP_FRONT_RIGHT       = 0x4000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_TOP_BACK_LEFT         = 0x8000u,
+    COEF_35,   // AUDIO_CHANNEL_OUT_TOP_BACK_CENTER       = 0x10000u,
+    0.f,       // AUDIO_CHANNEL_OUT_TOP_BACK_RIGHT        = 0x20000u,
+    COEF_61,   // AUDIO_CHANNEL_OUT_TOP_SIDE_LEFT         = 0x40000u,
+    0.f,       // AUDIO_CHANNEL_OUT_TOP_SIDE_RIGHT        = 0x80000u,
+    1.f,       // AUDIO_CHANNEL_OUT_BOTTOM_FRONT_LEFT     = 0x100000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_BOTTOM_FRONT_CENTER   = 0x200000u,
+    0.f, // AUDIO_CHANNEL_OUT_BOTTOM_FRONT_RIGHT    = 0x400000u,
+    0.f, // AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2       = 0x800000u,
+};
+
+constexpr inline float kScaleFromChannelIdxRight[] = {
+    0.f,       // AUDIO_CHANNEL_OUT_FRONT_LEFT            = 0x1u,
+    1.f,       // AUDIO_CHANNEL_OUT_FRONT_RIGHT           = 0x2u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_FRONT_CENTER          = 0x4u,
+    0.5f,      // AUDIO_CHANNEL_OUT_LOW_FREQUENCY         = 0x8u,
+    0.f,       // AUDIO_CHANNEL_OUT_BACK_LEFT             = 0x10u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_BACK_RIGHT            = 0x20u,
+    COEF_25,   // AUDIO_CHANNEL_OUT_FRONT_LEFT_OF_CENTER  = 0x40u,
+    COEF_61,   // AUDIO_CHANNEL_OUT_FRONT_RIGHT_OF_CENTER = 0x80u,
+    0.5f,      // AUDIO_CHANNEL_OUT_BACK_CENTER           = 0x100u,
+    0.f,       // AUDIO_CHANNEL_OUT_SIDE_LEFT             = 0x200u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_SIDE_RIGHT            = 0x400u,
+    COEF_36,   // AUDIO_CHANNEL_OUT_TOP_CENTER            = 0x800u,
+    0.f,       // AUDIO_CHANNEL_OUT_TOP_FRONT_LEFT        = 0x1000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_TOP_FRONT_CENTER      = 0x2000u,
+    1.f,       // AUDIO_CHANNEL_OUT_TOP_FRONT_RIGHT       = 0x4000u,
+    0.f,       // AUDIO_CHANNEL_OUT_TOP_BACK_LEFT         = 0x8000u,
+    COEF_35,   // AUDIO_CHANNEL_OUT_TOP_BACK_CENTER       = 0x10000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_TOP_BACK_RIGHT        = 0x20000u,
+    0.f,       // AUDIO_CHANNEL_OUT_TOP_SIDE_LEFT         = 0x40000u,
+    COEF_61,   // AUDIO_CHANNEL_OUT_TOP_SIDE_RIGHT        = 0x80000u,
+    0.f,       // AUDIO_CHANNEL_OUT_BOTTOM_FRONT_LEFT     = 0x100000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_BOTTOM_FRONT_CENTER   = 0x200000u,
+    1.f,       // AUDIO_CHANNEL_OUT_BOTTOM_FRONT_RIGHT    = 0x400000u,
+    M_SQRT1_2, // AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2       = 0x800000u,
+};
 
 // Downmix doesn't change with sample rate
 static constexpr size_t kSampleRates[] = {
@@ -93,8 +147,8 @@
     void testBalance(int sampleRate, audio_channel_mask_t channelMask) {
         using namespace ::android::audio_utils::channels;
 
-        size_t frames = 100;
-        unsigned outChannels = 2;
+        size_t frames = 100; // set to an even number (2, 4, 6 ... ) stream alternates +1, -1.
+        constexpr unsigned outChannels = 2;
         unsigned inChannels = audio_channel_count_from_out_mask(channelMask);
         std::vector<float> input(frames * inChannels);
         std::vector<float> output(frames * outChannels);
@@ -119,7 +173,7 @@
 
             auto stats = channelStatistics(output, 2 /* channels */);
             // printf("power: %s %s\n", stats[0].toString().c_str(), stats[1].toString().c_str());
-            double power[2] = { stats[0].getVariance(), stats[1].getVariance() };
+            double power[2] = { stats[0].getPopVariance(), stats[1].getPopVariance() };
 
             // Check symmetric power for pair channels on exchange of left/right position.
             // to do this, we save previous power measurements.
@@ -130,28 +184,39 @@
             savedPower[index][0] = power[0];
             savedPower[index][1] = power[1];
 
-            // Confirm exactly the mix amount prescribed by the existing downmix effect.
-            // For future changes to the downmix effect, the nearness needs to be relaxed
-            // to compare behavior S or earlier.
-            if ((channelBit & kConsideredChannels) == 0) {
-                // for channels not considered, expect 0 power for legacy downmix
-                EXPECT_EQ(0.f, power[0]);
-                EXPECT_EQ(0.f, power[1]);
-                continue;
-            }
-            constexpr float POWER_TOLERANCE = 0.01;  // for variance sum error.
+            constexpr float POWER_TOLERANCE = 0.001;
+            const float expectedPower =
+                    kScaleFromChannelIdxLeft[index] * kScaleFromChannelIdxLeft[index]
+                    + kScaleFromChannelIdxRight[index] * kScaleFromChannelIdxRight[index];
+            EXPECT_NEAR(expectedPower, power[0] + power[1], POWER_TOLERANCE);
             switch (side) {
             case AUDIO_GEOMETRY_SIDE_LEFT:
-                EXPECT_NEAR(0.25f, power[0], POWER_TOLERANCE);
+                if (channelBit == AUDIO_CHANNEL_OUT_FRONT_LEFT_OF_CENTER) {
+                    break;
+                }
                 EXPECT_EQ(0.f, power[1]);
                 break;
             case AUDIO_GEOMETRY_SIDE_RIGHT:
+                if (channelBit == AUDIO_CHANNEL_OUT_FRONT_RIGHT_OF_CENTER) {
+                    break;
+                }
                 EXPECT_EQ(0.f, power[0]);
-                EXPECT_NEAR(0.25f, power[1], POWER_TOLERANCE);
                 break;
             case AUDIO_GEOMETRY_SIDE_CENTER:
-                EXPECT_NEAR(0.125f, power[0], POWER_TOLERANCE);
-                EXPECT_NEAR(0.125f, power[1], POWER_TOLERANCE);
+                if (channelBit == AUDIO_CHANNEL_OUT_LOW_FREQUENCY) {
+                    if (channelMask & AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2) {
+                        EXPECT_EQ(0.f, power[1]);
+                        break;
+                    } else {
+                        EXPECT_NEAR_EPSILON(power[0], power[1]); // always true
+                        EXPECT_NEAR(expectedPower, power[0] + power[1], POWER_TOLERANCE);
+                        break;
+                    }
+                } else if (channelBit == AUDIO_CHANNEL_OUT_LOW_FREQUENCY_2) {
+                    EXPECT_EQ(0.f, power[0]);
+                    EXPECT_NEAR(expectedPower, power[1], POWER_TOLERANCE);
+                    break;
+                }
                 EXPECT_NEAR_EPSILON(power[0], power[1]);
                 break;
             }
@@ -244,10 +309,11 @@
         ::testing::Combine(
                 ::testing::Range(0, (int)std::size(kSampleRates)),
                 ::testing::Range(0, (int)std::size(kChannelPositionMasks))
-                ));
-
-int main(int argc, /* const */ char** argv) {
-    ::testing::InitGoogleTest(&argc, argv);
-    int status = RUN_ALL_TESTS();
-    return status;
-}
+                ),
+        [](const testing::TestParamInfo<DownmixTest::ParamType>& info) {
+            const int index = std::get<1>(info.param);
+            const audio_channel_mask_t channelMask = kChannelPositionMasks[index];
+            const std::string name = std::string(audio_channel_out_mask_to_string(channelMask))
+                + "_" + std::to_string(std::get<0>(info.param)) + "_" + std::to_string(index);
+            return name;
+        });
diff --git a/media/libeffects/hapticgenerator/Android.bp b/media/libeffects/hapticgenerator/Android.bp
index a660957..03ce329 100644
--- a/media/libeffects/hapticgenerator/Android.bp
+++ b/media/libeffects/hapticgenerator/Android.bp
@@ -45,6 +45,7 @@
 
     shared_libs: [
         "libaudioutils",
+        "libbase",
         "libbinder",
         "liblog",
         "libutils",
diff --git a/media/libeffects/hapticgenerator/EffectHapticGenerator.cpp b/media/libeffects/hapticgenerator/EffectHapticGenerator.cpp
index 65a20a7..3137e13 100644
--- a/media/libeffects/hapticgenerator/EffectHapticGenerator.cpp
+++ b/media/libeffects/hapticgenerator/EffectHapticGenerator.cpp
@@ -22,12 +22,15 @@
 
 #include <algorithm>
 #include <memory>
+#include <string>
 #include <utility>
 
 #include <errno.h>
 #include <inttypes.h>
 #include <math.h>
 
+#include <android-base/parsedouble.h>
+#include <android-base/properties.h>
 #include <audio_effects/effect_hapticgenerator.h>
 #include <audio_utils/format.h>
 #include <system/audio.h>
@@ -35,6 +38,7 @@
 static constexpr float DEFAULT_RESONANT_FREQUENCY = 150.0f;
 static constexpr float DEFAULT_BSF_ZERO_Q = 8.0f;
 static constexpr float DEFAULT_BSF_POLE_Q = 4.0f;
+static constexpr float DEFAULT_DISTORTION_OUTPUT_GAIN = 1.5f;
 
 // This is the only symbol that needs to be exported
 __attribute__ ((visibility ("default")))
@@ -81,6 +85,15 @@
 
 namespace {
 
+float getFloatProperty(const std::string& key, float defaultValue) {
+    float result;
+    std::string value = android::base::GetProperty(key, "");
+    if (!value.empty() && android::base::ParseFloat(value, &result)) {
+        return result;
+    }
+    return defaultValue;
+}
+
 int HapticGenerator_Init(struct HapticGeneratorContext *context) {
     context->itfe = &gHapticGeneratorInterface;
 
@@ -114,7 +127,9 @@
     context->param.distortionCornerFrequency = 300.0f;
     context->param.distortionInputGain = 0.3f;
     context->param.distortionCubeThreshold = 0.1f;
-    context->param.distortionOutputGain = 1.5f;
+    context->param.distortionOutputGain = getFloatProperty(
+            "vendor.audio.hapticgenerator.distortion.output.gain", DEFAULT_DISTORTION_OUTPUT_GAIN);
+    ALOGD("Using distortion output gain as %f", context->param.distortionOutputGain);
 
     context->state = HAPTICGENERATOR_STATE_INITIALIZED;
     return 0;
@@ -287,15 +302,17 @@
         break;
     }
     case HG_PARAM_VIBRATOR_INFO: {
-        if (value == nullptr || size != 2 * sizeof(float)) {
+        if (value == nullptr || size != 3 * sizeof(float)) {
             return -EINVAL;
         }
         const float resonantFrequency = *(float*) value;
         const float qFactor = *((float *) value + 1);
+        const float maxAmplitude = *((float *) value + 2);
         context->param.resonantFrequency =
                 isnan(resonantFrequency) ? DEFAULT_RESONANT_FREQUENCY : resonantFrequency;
         context->param.bsfZeroQ = isnan(qFactor) ? DEFAULT_BSF_POLE_Q : qFactor;
         context->param.bsfPoleQ = context->param.bsfZeroQ / 2.0f;
+        context->param.maxHapticAmplitude = maxAmplitude;
 
         if (context->processorsRecord.bpf != nullptr) {
             context->processorsRecord.bpf->setCoefficients(
@@ -448,7 +465,8 @@
     float* hapticOutBuffer = HapticGenerator_runProcessingChain(
             context->processingChain, context->inputBuffer.data(),
             context->outputBuffer.data(), inBuffer->frameCount);
-    os::scaleHapticData(hapticOutBuffer, hapticSampleCount, context->param.maxHapticIntensity);
+    os::scaleHapticData(hapticOutBuffer, hapticSampleCount, context->param.maxHapticIntensity,
+                        context->param.maxHapticAmplitude);
 
     // For haptic data, the haptic playback thread will copy the data from effect input buffer,
     // which contains haptic data at the end of the buffer, directly to sink buffer.
diff --git a/media/libeffects/hapticgenerator/EffectHapticGenerator.h b/media/libeffects/hapticgenerator/EffectHapticGenerator.h
index 96b744a..85e961f 100644
--- a/media/libeffects/hapticgenerator/EffectHapticGenerator.h
+++ b/media/libeffects/hapticgenerator/EffectHapticGenerator.h
@@ -51,6 +51,7 @@
     // A map from track id to haptic intensity.
     std::map<int, os::HapticScale> id2Intensity;
     os::HapticScale maxHapticIntensity; // max intensity will be used to scale haptic data.
+    float maxHapticAmplitude; // max amplitude will be used to limit haptic data absolute values.
 
     float resonantFrequency;
     float bpfQ;
diff --git a/media/libheadtracking/Android.bp b/media/libheadtracking/Android.bp
new file mode 100644
index 0000000..d134ca0
--- /dev/null
+++ b/media/libheadtracking/Android.bp
@@ -0,0 +1,79 @@
+package {
+    // See: http://go/android-license-faq
+    // A large-scale-change added 'default_applicable_licenses' to import
+    // all of the 'license_kinds' from "frameworks_av_license"
+    // to get the below license kinds:
+    //   SPDX-license-identifier-Apache-2.0
+    default_applicable_licenses: ["frameworks_av_license"],
+}
+
+cc_library {
+    name: "libheadtracking",
+    host_supported: true,
+    srcs: [
+      "HeadTrackingProcessor.cpp",
+      "ModeSelector.cpp",
+      "Pose.cpp",
+      "PoseDriftCompensator.cpp",
+      "PoseRateLimiter.cpp",
+      "QuaternionUtil.cpp",
+      "ScreenHeadFusion.cpp",
+      "Twist.cpp",
+    ],
+    export_include_dirs: [
+        "include",
+    ],
+    header_libs: [
+        "libeigen",
+    ],
+    export_header_lib_headers: [
+        "libeigen",
+    ],
+}
+
+cc_library {
+    name: "libheadtracking-binding",
+    srcs: [
+      "SensorPoseProvider.cpp",
+    ],
+    shared_libs: [
+        "libheadtracking",
+        "libandroid",
+        "liblog",
+        "libsensor",
+    ],
+    export_shared_lib_headers: [
+        "libheadtracking",
+    ],
+}
+
+cc_binary {
+    name: "SensorPoseProvider-example",
+    srcs: [
+        "SensorPoseProvider-example.cpp",
+    ],
+    shared_libs: [
+        "libandroid",
+        "libheadtracking",
+        "libheadtracking-binding",
+        "libsensor",
+        "libutils",
+    ],
+}
+
+cc_test_host {
+    name: "libheadtracking-test",
+    srcs: [
+        "HeadTrackingProcessor-test.cpp",
+        "ModeSelector-test.cpp",
+        "Pose-test.cpp",
+        "PoseDriftCompensator-test.cpp",
+        "PoseRateLimiter-test.cpp",
+        "QuaternionUtil-test.cpp",
+        "ScreenHeadFusion-test.cpp",
+        "Twist-test.cpp",
+    ],
+    shared_libs: [
+        "libheadtracking",
+    ],
+}
diff --git a/media/libheadtracking/HeadTrackingProcessor-test.cpp b/media/libheadtracking/HeadTrackingProcessor-test.cpp
new file mode 100644
index 0000000..299192f
--- /dev/null
+++ b/media/libheadtracking/HeadTrackingProcessor-test.cpp
@@ -0,0 +1,144 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "media/HeadTrackingProcessor.h"
+
+#include <gtest/gtest.h>
+
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+namespace android {
+namespace media {
+namespace {
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+using Options = HeadTrackingProcessor::Options;
+
+TEST(HeadTrackingProcessor, Initial) {
+    for (auto mode : {HeadTrackingMode::STATIC, HeadTrackingMode::WORLD_RELATIVE,
+                      HeadTrackingMode::SCREEN_RELATIVE}) {
+        std::unique_ptr<HeadTrackingProcessor> processor =
+                createHeadTrackingProcessor(Options{}, mode);
+        processor->calculate(0);
+        EXPECT_EQ(processor->getActualMode(), HeadTrackingMode::STATIC);
+        EXPECT_EQ(processor->getHeadToStagePose(), Pose3f());
+    }
+}
+
+TEST(HeadTrackingProcessor, BasicComposition) {
+    const Pose3f worldToHead{{1, 2, 3}, Quaternionf::UnitRandom()};
+    const Pose3f worldToScreen{{4, 5, 6}, Quaternionf::UnitRandom()};
+    const Pose3f screenToStage{{7, 8, 9}, Quaternionf::UnitRandom()};
+    const float physicalToLogical = M_PI_2;
+
+    std::unique_ptr<HeadTrackingProcessor> processor =
+            createHeadTrackingProcessor(Options{}, HeadTrackingMode::SCREEN_RELATIVE);
+
+    // Establish a baseline for the drift compensators.
+    processor->setWorldToHeadPose(0, Pose3f(), Twist3f());
+    processor->setWorldToScreenPose(0, Pose3f());
+
+    processor->setDisplayOrientation(physicalToLogical);
+    processor->setWorldToHeadPose(0, worldToHead, Twist3f());
+    processor->setWorldToScreenPose(0, worldToScreen);
+    processor->setScreenToStagePose(screenToStage);
+    processor->calculate(0);
+    ASSERT_EQ(processor->getActualMode(), HeadTrackingMode::SCREEN_RELATIVE);
+    EXPECT_EQ(processor->getHeadToStagePose(), worldToHead.inverse() * worldToScreen *
+                                                       Pose3f(rotateY(-physicalToLogical)) *
+                                                       screenToStage);
+
+    processor->setDesiredMode(HeadTrackingMode::WORLD_RELATIVE);
+    processor->calculate(0);
+    ASSERT_EQ(processor->getActualMode(), HeadTrackingMode::WORLD_RELATIVE);
+    EXPECT_EQ(processor->getHeadToStagePose(), worldToHead.inverse() * screenToStage);
+
+    processor->setDesiredMode(HeadTrackingMode::STATIC);
+    processor->calculate(0);
+    ASSERT_EQ(processor->getActualMode(), HeadTrackingMode::STATIC);
+    EXPECT_EQ(processor->getHeadToStagePose(), screenToStage);
+}
+
+TEST(HeadTrackingProcessor, Prediction) {
+    const Pose3f worldToHead{{1, 2, 3}, Quaternionf::UnitRandom()};
+    const Twist3f headTwist{{4, 5, 6}, quaternionToRotationVector(Quaternionf::UnitRandom()) / 10};
+    const Pose3f worldToScreen{{4, 5, 6}, Quaternionf::UnitRandom()};
+
+    std::unique_ptr<HeadTrackingProcessor> processor = createHeadTrackingProcessor(
+            Options{.predictionDuration = 2.f}, HeadTrackingMode::WORLD_RELATIVE);
+
+    // Establish a baseline for the drift compensators.
+    processor->setWorldToHeadPose(0, Pose3f(), Twist3f());
+    processor->setWorldToScreenPose(0, Pose3f());
+
+    processor->setWorldToHeadPose(0, worldToHead, headTwist);
+    processor->setWorldToScreenPose(0, worldToScreen);
+    processor->calculate(0);
+    ASSERT_EQ(processor->getActualMode(), HeadTrackingMode::WORLD_RELATIVE);
+    EXPECT_EQ(processor->getHeadToStagePose(), (worldToHead * integrate(headTwist, 2.f)).inverse());
+
+    processor->setDesiredMode(HeadTrackingMode::SCREEN_RELATIVE);
+    processor->calculate(0);
+    ASSERT_EQ(processor->getActualMode(), HeadTrackingMode::SCREEN_RELATIVE);
+    EXPECT_EQ(processor->getHeadToStagePose(),
+              (worldToHead * integrate(headTwist, 2.f)).inverse() * worldToScreen);
+
+    processor->setDesiredMode(HeadTrackingMode::STATIC);
+    processor->calculate(0);
+    ASSERT_EQ(processor->getActualMode(), HeadTrackingMode::STATIC);
+    EXPECT_EQ(processor->getHeadToStagePose(), Pose3f());
+}
+
+TEST(HeadTrackingProcessor, SmoothModeSwitch) {
+    const Pose3f targetHeadToWorld = Pose3f({4, 0, 0}, rotateZ(M_PI / 2));
+
+    std::unique_ptr<HeadTrackingProcessor> processor = createHeadTrackingProcessor(
+            Options{.maxTranslationalVelocity = 1}, HeadTrackingMode::STATIC);
+
+    // Establish a baseline for the drift compensators.
+    processor->setWorldToHeadPose(0, Pose3f(), Twist3f());
+    processor->setWorldToScreenPose(0, Pose3f());
+
+    processor->calculate(0);
+
+    processor->setDesiredMode(HeadTrackingMode::WORLD_RELATIVE);
+    processor->setWorldToHeadPose(0, targetHeadToWorld.inverse(), Twist3f());
+
+    // We're expecting a gradual move to the target.
+    processor->calculate(0);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, processor->getActualMode());
+    EXPECT_EQ(processor->getHeadToStagePose(), Pose3f());
+
+    processor->calculate(2);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, processor->getActualMode());
+    EXPECT_EQ(processor->getHeadToStagePose(), Pose3f({2, 0, 0}, rotateZ(M_PI / 4)));
+
+    processor->calculate(4);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, processor->getActualMode());
+    EXPECT_EQ(processor->getHeadToStagePose(), targetHeadToWorld);
+
+    // Now that we've reached the target, we should no longer be rate limiting.
+    processor->setWorldToHeadPose(4, Pose3f(), Twist3f());
+    processor->calculate(5);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, processor->getActualMode());
+    EXPECT_EQ(processor->getHeadToStagePose(), Pose3f());
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/HeadTrackingProcessor.cpp b/media/libheadtracking/HeadTrackingProcessor.cpp
new file mode 100644
index 0000000..47f7cf0
--- /dev/null
+++ b/media/libheadtracking/HeadTrackingProcessor.cpp
@@ -0,0 +1,156 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"){}
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "media/HeadTrackingProcessor.h"
+
+#include "ModeSelector.h"
+#include "PoseDriftCompensator.h"
+#include "QuaternionUtil.h"
+#include "ScreenHeadFusion.h"
+
+namespace android {
+namespace media {
+namespace {
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+class HeadTrackingProcessorImpl : public HeadTrackingProcessor {
+  public:
+    HeadTrackingProcessorImpl(const Options& options, HeadTrackingMode initialMode)
+        : mOptions(options),
+          mHeadPoseDriftCompensator(PoseDriftCompensator::Options{
+                  .translationalDriftTimeConstant = options.translationalDriftTimeConstant,
+                  .rotationalDriftTimeConstant = options.rotationalDriftTimeConstant,
+          }),
+          mScreenPoseDriftCompensator(PoseDriftCompensator::Options{
+                  .translationalDriftTimeConstant = options.translationalDriftTimeConstant,
+                  .rotationalDriftTimeConstant = options.rotationalDriftTimeConstant,
+          }),
+          mModeSelector(ModeSelector::Options{.freshnessTimeout = options.freshnessTimeout},
+                        initialMode),
+          mRateLimiter(PoseRateLimiter::Options{
+                  .maxTranslationalVelocity = options.maxTranslationalVelocity,
+                  .maxRotationalVelocity = options.maxRotationalVelocity}) {}
+
+    void setDesiredMode(HeadTrackingMode mode) override { mModeSelector.setDesiredMode(mode); }
+
+    void setWorldToHeadPose(int64_t timestamp, const Pose3f& worldToHead,
+                            const Twist3f& headTwist) override {
+        Pose3f predictedWorldToHead =
+                worldToHead * integrate(headTwist, mOptions.predictionDuration);
+        mHeadPoseDriftCompensator.setInput(timestamp, predictedWorldToHead);
+        mWorldToHeadTimestamp = timestamp;
+    }
+
+    void setWorldToScreenPose(int64_t timestamp, const Pose3f& worldToScreen) override {
+        if (mPhysicalToLogicalAngle != mPendingPhysicalToLogicalAngle) {
+            // We're introducing an artificial discontinuity. Enable the rate limiter.
+            mRateLimiter.enable();
+            mPhysicalToLogicalAngle = mPendingPhysicalToLogicalAngle;
+        }
+
+        mScreenPoseDriftCompensator.setInput(
+                timestamp, worldToScreen * Pose3f(rotateY(-mPhysicalToLogicalAngle)));
+        mWorldToScreenTimestamp = timestamp;
+    }
+
+    void setScreenToStagePose(const Pose3f& screenToStage) override {
+        mModeSelector.setScreenToStagePose(screenToStage);
+    }
+
+    void setDisplayOrientation(float physicalToLogicalAngle) override {
+        mPendingPhysicalToLogicalAngle = physicalToLogicalAngle;
+    }
+
+    void calculate(int64_t timestamp) override {
+        if (mWorldToHeadTimestamp.has_value()) {
+            const Pose3f worldToHead = mHeadPoseDriftCompensator.getOutput();
+            mScreenHeadFusion.setWorldToHeadPose(mWorldToHeadTimestamp.value(), worldToHead);
+            mModeSelector.setWorldToHeadPose(mWorldToHeadTimestamp.value(), worldToHead);
+        }
+
+        if (mWorldToScreenTimestamp.has_value()) {
+            const Pose3f worldToLogicalScreen = mScreenPoseDriftCompensator.getOutput();
+            mScreenHeadFusion.setWorldToScreenPose(mWorldToScreenTimestamp.value(),
+                                                   worldToLogicalScreen);
+        }
+
+        auto maybeScreenToHead = mScreenHeadFusion.calculate();
+        if (maybeScreenToHead.has_value()) {
+            mModeSelector.setScreenToHeadPose(maybeScreenToHead->timestamp,
+                                              maybeScreenToHead->pose);
+        } else {
+            mModeSelector.setScreenToHeadPose(timestamp, std::nullopt);
+        }
+
+        HeadTrackingMode prevMode = mModeSelector.getActualMode();
+        mModeSelector.calculate(timestamp);
+        if (mModeSelector.getActualMode() != prevMode) {
+            // Mode has changed, enable rate limiting.
+            mRateLimiter.enable();
+        }
+        mRateLimiter.setTarget(mModeSelector.getHeadToStagePose());
+        mHeadToStagePose = mRateLimiter.calculatePose(timestamp);
+    }
+
+    Pose3f getHeadToStagePose() const override { return mHeadToStagePose; }
+
+    HeadTrackingMode getActualMode() const override { return mModeSelector.getActualMode(); }
+
+    void recenter(bool recenterHead, bool recenterScreen) override {
+        if (recenterHead) {
+            mHeadPoseDriftCompensator.recenter();
+        }
+        if (recenterScreen) {
+            mScreenPoseDriftCompensator.recenter();
+        }
+
+        // If a sensor being recentered is included in the current mode, apply rate limiting to
+        // avoid discontinuities.
+        HeadTrackingMode mode = mModeSelector.getActualMode();
+        if ((recenterHead && (mode == HeadTrackingMode::WORLD_RELATIVE ||
+                              mode == HeadTrackingMode::SCREEN_RELATIVE)) ||
+            (recenterScreen && mode == HeadTrackingMode::SCREEN_RELATIVE)) {
+            mRateLimiter.enable();
+        }
+    }
+
+  private:
+    const Options mOptions;
+    float mPhysicalToLogicalAngle = 0;
+    // We store the physical to logical angle as "pending" until the next world-to-screen sample it
+    // applies to arrives.
+    float mPendingPhysicalToLogicalAngle = 0;
+    std::optional<int64_t> mWorldToHeadTimestamp;
+    std::optional<int64_t> mWorldToScreenTimestamp;
+    Pose3f mHeadToStagePose;
+    PoseDriftCompensator mHeadPoseDriftCompensator;
+    PoseDriftCompensator mScreenPoseDriftCompensator;
+    ScreenHeadFusion mScreenHeadFusion;
+    ModeSelector mModeSelector;
+    PoseRateLimiter mRateLimiter;
+};
+
+}  // namespace
+
+std::unique_ptr<HeadTrackingProcessor> createHeadTrackingProcessor(
+        const HeadTrackingProcessor::Options& options, HeadTrackingMode initialMode) {
+    return std::make_unique<HeadTrackingProcessorImpl>(options, initialMode);
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/ModeSelector-test.cpp b/media/libheadtracking/ModeSelector-test.cpp
new file mode 100644
index 0000000..6247d84
--- /dev/null
+++ b/media/libheadtracking/ModeSelector-test.cpp
@@ -0,0 +1,149 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "ModeSelector.h"
+
+#include <gtest/gtest.h>
+
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+namespace android {
+namespace media {
+namespace {
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+TEST(ModeSelector, Initial) {
+    ModeSelector::Options options;
+    ModeSelector selector(options);
+
+    selector.calculate(0);
+    EXPECT_EQ(HeadTrackingMode::STATIC, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), Pose3f());
+}
+
+TEST(ModeSelector, InitialWorldRelative) {
+    const Pose3f worldToHead({1, 2, 3}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options;
+    ModeSelector selector(options, HeadTrackingMode::WORLD_RELATIVE);
+
+    selector.setWorldToHeadPose(0, worldToHead);
+    selector.calculate(0);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), worldToHead.inverse());
+}
+
+TEST(ModeSelector, InitialScreenRelative) {
+    const Pose3f screenToHead({1, 2, 3}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options;
+    ModeSelector selector(options, HeadTrackingMode::SCREEN_RELATIVE);
+
+    selector.setScreenToHeadPose(0, screenToHead);
+    selector.calculate(0);
+    EXPECT_EQ(HeadTrackingMode::SCREEN_RELATIVE, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), screenToHead.inverse());
+}
+
+TEST(ModeSelector, WorldRelative) {
+    const Pose3f worldToHead({1, 2, 3}, Quaternionf::UnitRandom());
+    const Pose3f screenToStage({4, 5, 6}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options;
+    ModeSelector selector(options);
+
+    selector.setScreenToStagePose(screenToStage);
+
+    selector.setDesiredMode(HeadTrackingMode::WORLD_RELATIVE);
+    selector.setWorldToHeadPose(0, worldToHead);
+    selector.calculate(0);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), worldToHead.inverse() * screenToStage);
+}
+
+TEST(ModeSelector, WorldRelativeStale) {
+    const Pose3f worldToHead({1, 2, 3}, Quaternionf::UnitRandom());
+    const Pose3f screenToStage({4, 5, 6}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options{.freshnessTimeout = 100};
+    ModeSelector selector(options);
+
+    selector.setScreenToStagePose(screenToStage);
+
+    selector.setDesiredMode(HeadTrackingMode::WORLD_RELATIVE);
+    selector.setWorldToHeadPose(0, worldToHead);
+    selector.calculate(101);
+    EXPECT_EQ(HeadTrackingMode::STATIC, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), screenToStage);
+}
+
+TEST(ModeSelector, ScreenRelative) {
+    const Pose3f screenToHead({1, 2, 3}, Quaternionf::UnitRandom());
+    const Pose3f screenToStage({4, 5, 6}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options;
+    ModeSelector selector(options);
+
+    selector.setScreenToStagePose(screenToStage);
+
+    selector.setDesiredMode(HeadTrackingMode::SCREEN_RELATIVE);
+    selector.setScreenToHeadPose(0, screenToHead);
+    selector.calculate(0);
+    EXPECT_EQ(HeadTrackingMode::SCREEN_RELATIVE, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), screenToHead.inverse() * screenToStage);
+}
+
+TEST(ModeSelector, ScreenRelativeStaleToWorldRelative) {
+    const Pose3f screenToHead({1, 2, 3}, Quaternionf::UnitRandom());
+    const Pose3f screenToStage({4, 5, 6}, Quaternionf::UnitRandom());
+    const Pose3f worldToHead({7, 8, 9}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options{.freshnessTimeout = 100};
+    ModeSelector selector(options);
+
+    selector.setScreenToStagePose(screenToStage);
+
+    selector.setDesiredMode(HeadTrackingMode::SCREEN_RELATIVE);
+    selector.setScreenToHeadPose(0, screenToHead);
+    selector.setWorldToHeadPose(50, worldToHead);
+    selector.calculate(101);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), worldToHead.inverse() * screenToStage);
+}
+
+TEST(ModeSelector, ScreenRelativeInvalidToWorldRelative) {
+    const Pose3f screenToStage({4, 5, 6}, Quaternionf::UnitRandom());
+    const Pose3f worldToHead({7, 8, 9}, Quaternionf::UnitRandom());
+
+    ModeSelector::Options options;
+    ModeSelector selector(options);
+
+    selector.setScreenToStagePose(screenToStage);
+
+    selector.setDesiredMode(HeadTrackingMode::SCREEN_RELATIVE);
+    selector.setScreenToHeadPose(50, std::nullopt);
+    selector.setWorldToHeadPose(50, worldToHead);
+    selector.calculate(101);
+    EXPECT_EQ(HeadTrackingMode::WORLD_RELATIVE, selector.getActualMode());
+    EXPECT_EQ(selector.getHeadToStagePose(), worldToHead.inverse() * screenToStage);
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/ModeSelector.cpp b/media/libheadtracking/ModeSelector.cpp
new file mode 100644
index 0000000..16e1712
--- /dev/null
+++ b/media/libheadtracking/ModeSelector.cpp
@@ -0,0 +1,96 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "ModeSelector.h"
+
+namespace android {
+namespace media {
+
+ModeSelector::ModeSelector(const Options& options, HeadTrackingMode initialMode)
+    : mOptions(options), mDesiredMode(initialMode), mActualMode(initialMode) {}
+
+void ModeSelector::setDesiredMode(HeadTrackingMode mode) {
+    mDesiredMode = mode;
+}
+
+void ModeSelector::setScreenToStagePose(const Pose3f& screenToStage) {
+    mScreenToStage = screenToStage;
+}
+
+void ModeSelector::setScreenToHeadPose(int64_t timestamp,
+                                       const std::optional<Pose3f>& screenToHead) {
+    mScreenToHead = screenToHead;
+    mScreenToHeadTimestamp = timestamp;
+}
+
+void ModeSelector::setWorldToHeadPose(int64_t timestamp, const Pose3f& worldToHead) {
+    mWorldToHead = worldToHead;
+    mWorldToHeadTimestamp = timestamp;
+}
+
+void ModeSelector::calculateActualMode(int64_t timestamp) {
+    bool isValidScreenToHead = mScreenToHead.has_value() &&
+                               timestamp - mScreenToHeadTimestamp < mOptions.freshnessTimeout;
+    bool isValidWorldToHead = mWorldToHead.has_value() &&
+                              timestamp - mWorldToHeadTimestamp < mOptions.freshnessTimeout;
+
+    HeadTrackingMode mode = mDesiredMode;
+
+    // Optional downgrade from screen-relative to world-relative.
+    if (mode == HeadTrackingMode::SCREEN_RELATIVE) {
+        if (!isValidScreenToHead) {
+            mode = HeadTrackingMode::WORLD_RELATIVE;
+        }
+    }
+
+    // Optional downgrade from world-relative to static.
+    if (mode == HeadTrackingMode::WORLD_RELATIVE) {
+        if (!isValidWorldToHead) {
+            mode = HeadTrackingMode::STATIC;
+        }
+    }
+
+    mActualMode = mode;
+}
+
+void ModeSelector::calculate(int64_t timestamp) {
+    calculateActualMode(timestamp);
+
+    switch (mActualMode) {
+        case HeadTrackingMode::STATIC:
+            mHeadToStage = mScreenToStage;
+            break;
+
+        case HeadTrackingMode::WORLD_RELATIVE:
+            mHeadToStage = mWorldToHead.value().inverse() * mScreenToStage;
+            break;
+
+        case HeadTrackingMode::SCREEN_RELATIVE:
+            mHeadToStage = mScreenToHead.value().inverse() * mScreenToStage;
+            break;
+    }
+}
+
+Pose3f ModeSelector::getHeadToStagePose() const {
+    return mHeadToStage;
+}
+
+HeadTrackingMode ModeSelector::getActualMode() const {
+    return mActualMode;
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/ModeSelector.h b/media/libheadtracking/ModeSelector.h
new file mode 100644
index 0000000..17a5142
--- /dev/null
+++ b/media/libheadtracking/ModeSelector.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <optional>
+
+#include "media/HeadTrackingMode.h"
+#include "media/Pose.h"
+
+#include "PoseRateLimiter.h"
+
+namespace android {
+namespace media {
+
+/**
+ * Head-tracking mode selector.
+ *
+ * This class is responsible for production of the determining pose for audio virtualization, based
+ * on a number of available sources and a selectable mode.
+ *
+ * Typical flow is:
+ * ModeSelector selector(...);
+ * while (...) {
+ *     // Set inputs.
+ *     selector.setFoo(...);
+ *     selector.setBar(...);
+ *
+ *     // Update outputs based on inputs.
+ *     selector.calculate(...);
+ *
+ *     // Get outputs.
+ *     Pose3f pose = selector.getHeadToStagePose();
+ * }
+ *
+ * This class is not thread-safe, but thread-compatible.
+ *
+ * For details on the frames of reference involved, their composition and the definitions to the
+ * different modes, refer to:
+ * go/immersive-audio-frames
+ *
+ * The actual mode may deviate from the desired mode in the following cases:
+ * - When we cannot get a valid and fresh estimate of the screen-to-head pose, we will fall back
+ *   from screen-relative to world-relative.
+ * - When we cannot get a fresh estimate of the world-to-head pose, we will fall back from
+ *   world-relative to static.
+ *
+ * All the timestamps used here are of arbitrary units and origin. They just need to be consistent
+ * between all the calls and with the Options provided for determining freshness and rate limiting.
+ */
+class ModeSelector {
+  public:
+    struct Options {
+        int64_t freshnessTimeout = std::numeric_limits<int64_t>::max();
+    };
+
+    ModeSelector(const Options& options, HeadTrackingMode initialMode = HeadTrackingMode::STATIC);
+
+    /** Sets the desired head-tracking mode. */
+    void setDesiredMode(HeadTrackingMode mode);
+
+    /**
+     * Set the screen-to-stage pose, used in all modes.
+     */
+    void setScreenToStagePose(const Pose3f& screenToStage);
+
+    /**
+     * Set the screen-to-head pose, used in screen-relative mode.
+     * The timestamp needs to reflect how fresh the sample is (not necessarily which point in time
+     * it applies to). nullopt can be used if it is determined that the listener is not in front of
+     * the screen.
+     */
+    void setScreenToHeadPose(int64_t timestamp, const std::optional<Pose3f>& screenToHead);
+
+    /**
+     * Set the world-to-head pose, used in world-relative mode.
+     * The timestamp needs to reflect how fresh the sample is (not necessarily which point in time
+     * it applies to).
+     */
+    void setWorldToHeadPose(int64_t timestamp, const Pose3f& worldToHead);
+
+    /**
+     * Process all the previous inputs and update the outputs.
+     */
+    void calculate(int64_t timestamp);
+
+    /**
+     * Get the aggregate head-to-stage pose (primary output of this module).
+     */
+    Pose3f getHeadToStagePose() const;
+
+    /**
+     * Get the actual head-tracking mode (which may deviate from the desired one as mentioned in the
+     * class documentation above).
+     */
+    HeadTrackingMode getActualMode() const;
+
+  private:
+    const Options mOptions;
+
+    HeadTrackingMode mDesiredMode;
+    Pose3f mScreenToStage;
+    std::optional<Pose3f> mScreenToHead;
+    int64_t mScreenToHeadTimestamp;
+    std::optional<Pose3f> mWorldToHead;
+    int64_t mWorldToHeadTimestamp;
+
+    HeadTrackingMode mActualMode;
+    Pose3f mHeadToStage;
+
+    void calculateActualMode(int64_t timestamp);
+};
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/OWNERS b/media/libheadtracking/OWNERS
new file mode 100644
index 0000000..e5d0370
--- /dev/null
+++ b/media/libheadtracking/OWNERS
@@ -0,0 +1,2 @@
+ytai@google.com
+elaurent@google.com
diff --git a/media/libheadtracking/Pose-test.cpp b/media/libheadtracking/Pose-test.cpp
new file mode 100644
index 0000000..a9e18ce
--- /dev/null
+++ b/media/libheadtracking/Pose-test.cpp
@@ -0,0 +1,138 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "media/Pose.h"
+
+#include <gtest/gtest.h>
+
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+using android::media::Pose3f;
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+namespace android {
+namespace media {
+namespace {
+
+TEST(Pose, CtorDefault) {
+    Pose3f pose;
+    EXPECT_EQ(pose.translation(), Vector3f::Zero());
+    EXPECT_EQ(pose.rotation(), Quaternionf::Identity());
+}
+
+TEST(Pose, CtorRotation) {
+    Quaternionf rot = Quaternionf::UnitRandom();
+    Pose3f pose(rot);
+    EXPECT_EQ(pose.translation(), Vector3f::Zero());
+    EXPECT_EQ(pose.rotation(), rot);
+}
+
+TEST(Pose, CtorTranslation) {
+    Vector3f trans{1, 2, 3};
+    Pose3f pose(trans);
+    EXPECT_EQ(pose.translation(), trans);
+    EXPECT_EQ(pose.rotation(), Quaternionf::Identity());
+}
+
+TEST(Pose, CtorTranslationRotation) {
+    Quaternionf rot = Quaternionf::UnitRandom();
+    Vector3f trans{1, 2, 3};
+    Pose3f pose(trans, rot);
+    EXPECT_EQ(pose.translation(), trans);
+    EXPECT_EQ(pose.rotation(), rot);
+}
+
+TEST(Pose, Inverse) {
+    Pose3f pose({1, 2, 3}, Quaternionf::UnitRandom());
+    EXPECT_EQ(pose.inverse() * pose, Pose3f());
+    EXPECT_EQ(pose * pose.inverse(), Pose3f());
+}
+
+TEST(Pose, IsApprox) {
+    constexpr float eps = std::numeric_limits<float>::epsilon();
+
+    EXPECT_EQ(Pose3f({1, 2, 3}, rotationVectorToQuaternion({4, 5, 6})),
+              Pose3f({1 + eps, 2 + eps, 3 + eps},
+                     rotationVectorToQuaternion({4 + eps, 5 + eps, 6 + eps})));
+
+    EXPECT_NE(Pose3f({1, 2, 3}, rotationVectorToQuaternion({4, 5, 6})),
+              Pose3f({1.01, 2, 3}, rotationVectorToQuaternion({4, 5, 6})));
+
+    EXPECT_NE(Pose3f({1, 2, 3}, rotationVectorToQuaternion({4, 5, 6})),
+              Pose3f({1, 2, 3}, rotationVectorToQuaternion({4.01, 5, 6})));
+}
+
+TEST(Pose, Compose) {
+    Pose3f p1({1, 2, 3}, rotateZ(M_PI_2));
+    Pose3f p2({4, 5, 6}, rotateX(M_PI_2));
+    Pose3f p3({-4, 6, 9}, p1.rotation() * p2.rotation());
+    EXPECT_EQ(p1 * p2, p3);
+}
+
+TEST(Pose, MoveWithRateLimit_NoLimit) {
+    Pose3f from({1, 1, 1}, Quaternionf::Identity());
+    Pose3f to({1, 1, 2}, rotateZ(M_PI_2));
+    auto result = moveWithRateLimit(from, to, 1, 10, 10);
+    EXPECT_EQ(std::get<0>(result), to);
+    EXPECT_FALSE(std::get<1>(result));
+}
+
+TEST(Pose, MoveWithRateLimit_TranslationLimit) {
+    Pose3f from({1, 1, 1}, Quaternionf::Identity());
+    Pose3f to({1, 1, 2}, rotateZ(M_PI_2));
+    auto result = moveWithRateLimit(from, to, 1, 0.5f, 10);
+    Pose3f expected({1, 1, 1.5f}, rotateZ(M_PI_4));
+    EXPECT_EQ(std::get<0>(result), expected);
+    EXPECT_TRUE(std::get<1>(result));
+}
+
+TEST(Pose, MoveWithRateLimit_RotationLimit) {
+    Pose3f from({1, 1, 1}, Quaternionf::Identity());
+    Pose3f to({1, 1, 2}, rotateZ(M_PI_2));
+    auto result = moveWithRateLimit(from, to, 1, 10, M_PI_4);
+    Pose3f expected({1, 1, 1.5f}, rotateZ(M_PI_4));
+    EXPECT_EQ(std::get<0>(result), expected);
+    EXPECT_TRUE(std::get<1>(result));
+}
+
+TEST(Pose, FloatVectorRoundTrip1) {
+    // Rotation vector magnitude must be less than Pi.
+    std::vector<float> vec = { 1, 2, 3, 0.4, 0.5, 0.6};
+    std::optional<Pose3f> pose = Pose3f::fromVector(vec);
+    ASSERT_TRUE(pose.has_value());
+    std::vector<float> reconstructed = pose->toVector();
+    EXPECT_EQ(vec, reconstructed);
+}
+
+TEST(Pose, FloatVectorRoundTrip2) {
+    Pose3f pose({1, 2, 3}, Quaternionf::UnitRandom());
+    std::vector<float> vec = pose.toVector();
+    std::optional<Pose3f> reconstructed = Pose3f::fromVector(vec);
+    ASSERT_TRUE(reconstructed.has_value());
+    EXPECT_EQ(pose, reconstructed.value());
+}
+
+TEST(Pose, FloatVectorInvalid) {
+    EXPECT_FALSE(Pose3f::fromVector({}).has_value());
+    EXPECT_FALSE(Pose3f::fromVector({1, 2, 3, 4, 5}).has_value());
+    EXPECT_FALSE(Pose3f::fromVector({1, 2, 3, 4, 5, 6, 7}).has_value());
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/Pose.cpp b/media/libheadtracking/Pose.cpp
new file mode 100644
index 0000000..47241ce
--- /dev/null
+++ b/media/libheadtracking/Pose.cpp
@@ -0,0 +1,69 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "media/Pose.h"
+#include "media/Twist.h"
+#include "QuaternionUtil.h"
+
+namespace android {
+namespace media {
+
+using Eigen::Vector3f;
+
+std::optional<Pose3f> Pose3f::fromVector(const std::vector<float>& vec) {
+    if (vec.size() != 6) {
+        return std::nullopt;
+    }
+    return Pose3f({vec[0], vec[1], vec[2]}, rotationVectorToQuaternion({vec[3], vec[4], vec[5]}));
+}
+
+std::vector<float> Pose3f::toVector() const {
+    Eigen::Vector3f rot = quaternionToRotationVector(mRotation);
+    return {mTranslation[0], mTranslation[1], mTranslation[2], rot[0], rot[1], rot[2]};
+}
+
+std::tuple<Pose3f, bool> moveWithRateLimit(const Pose3f& from, const Pose3f& to, float t,
+                                           float maxTranslationalVelocity,
+                                           float maxRotationalVelocity) {
+    // Never rate limit if both limits are set to infinity.
+    if (isinf(maxTranslationalVelocity) && isinf(maxRotationalVelocity)) {
+        return {to, false};
+    }
+    // Always rate limit if t is 0 (required to avoid division by 0).
+    if (t == 0) {
+        return {from, true};
+    }
+
+    Pose3f fromToTo = from.inverse() * to;
+    Twist3f twist = differentiate(fromToTo, t);
+    float angularRotationalRatio = twist.scalarRotationalVelocity() / maxRotationalVelocity;
+    float translationalVelocityRatio =
+            twist.scalarTranslationalVelocity() / maxTranslationalVelocity;
+    float maxRatio = std::max(angularRotationalRatio, translationalVelocityRatio);
+    if (maxRatio <= 1) {
+        return {to, false};
+    }
+    return {from * integrate(twist, t / maxRatio), true};
+}
+
+std::ostream& operator<<(std::ostream& os, const Pose3f& pose) {
+    os << "translation: " << pose.translation().transpose()
+       << " quaternion: " << pose.rotation().coeffs().transpose();
+    return os;
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/PoseDriftCompensator-test.cpp b/media/libheadtracking/PoseDriftCompensator-test.cpp
new file mode 100644
index 0000000..df0a05f
--- /dev/null
+++ b/media/libheadtracking/PoseDriftCompensator-test.cpp
@@ -0,0 +1,146 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <gtest/gtest.h>
+#include <cmath>
+
+#include "PoseDriftCompensator.h"
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+namespace android {
+namespace media {
+namespace {
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+using Options = PoseDriftCompensator::Options;
+
+TEST(PoseDriftCompensator, Initial) {
+    PoseDriftCompensator comp(Options{});
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+}
+
+TEST(PoseDriftCompensator, NoDrift) {
+    Pose3f pose1({1, 2, 3}, Quaternionf::UnitRandom());
+    Pose3f pose2({4, 5, 6}, Quaternionf::UnitRandom());
+    PoseDriftCompensator comp(Options{});
+
+    // First pose sets the baseline.
+    comp.setInput(1000, pose1);
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(2000, pose2);
+    EXPECT_EQ(comp.getOutput(), pose1.inverse() * pose2);
+
+    // Recentering resets the baseline.
+    comp.recenter();
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(3000, pose1);
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(4000, pose2);
+    EXPECT_EQ(comp.getOutput(), pose1.inverse() * pose2);
+}
+
+TEST(PoseDriftCompensator, NoDriftZeroTime) {
+    Pose3f pose1({1, 2, 3}, Quaternionf::UnitRandom());
+    Pose3f pose2({4, 5, 6}, Quaternionf::UnitRandom());
+    PoseDriftCompensator comp(Options{});
+
+    comp.setInput(1000, pose1);
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(1000, pose2);
+    EXPECT_EQ(comp.getOutput(), pose1.inverse() * pose2);
+
+    comp.recenter();
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(1000, pose1);
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(1000, pose2);
+    EXPECT_EQ(comp.getOutput(), pose1.inverse() * pose2);
+}
+
+TEST(PoseDriftCompensator, Asymptotic) {
+    Pose3f pose({1, 2, 3}, Quaternionf::UnitRandom());
+
+    PoseDriftCompensator comp(
+            Options{.translationalDriftTimeConstant = 1, .rotationalDriftTimeConstant = 1});
+
+    // Set the same pose for a long time.
+    for (int64_t t = 0; t < 1000; ++t) {
+        comp.setInput(t, pose);
+    }
+
+    // Output would have faded to approx. identity.
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+}
+
+TEST(PoseDriftCompensator, Fast) {
+    Pose3f pose1({1, 2, 3}, Quaternionf::UnitRandom());
+    Pose3f pose2({4, 5, 6}, Quaternionf::UnitRandom());
+    PoseDriftCompensator comp(
+            Options{.translationalDriftTimeConstant = 1e7, .rotationalDriftTimeConstant = 1e7});
+
+    comp.setInput(0, pose1);
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(1, pose2);
+    EXPECT_EQ(comp.getOutput(), pose1.inverse() * pose2);
+
+    comp.recenter();
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(2, pose1);
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+
+    comp.setInput(3, pose2);
+    EXPECT_EQ(comp.getOutput(), pose1.inverse() * pose2);
+}
+
+TEST(PoseDriftCompensator, Drift) {
+    Pose3f pose1({1, 2, 3}, rotateZ(-M_PI * 3 / 4));
+    PoseDriftCompensator comp(
+            Options{.translationalDriftTimeConstant = 500, .rotationalDriftTimeConstant = 1000});
+
+    // Establish a baseline.
+    comp.setInput(1000, Pose3f());
+
+    // Initial pose is used as is.
+    comp.setInput(1000, pose1);
+    EXPECT_EQ(comp.getOutput(), pose1);
+
+    // After 1000 ticks, our rotation should be exp(-1) and translation exp(-2) from identity.
+    comp.setInput(2000, pose1);
+    EXPECT_EQ(comp.getOutput(),
+              Pose3f(Vector3f{1, 2, 3} * std::expf(-2), rotateZ(-M_PI * 3 / 4 * std::expf(-1))));
+
+    // As long as the input stays the same, we'll continue to advance towards identity.
+    comp.setInput(3000, pose1);
+    EXPECT_EQ(comp.getOutput(),
+              Pose3f(Vector3f{1, 2, 3} * std::expf(-4), rotateZ(-M_PI * 3 / 4 * std::expf(-2))));
+
+    comp.recenter();
+    EXPECT_EQ(comp.getOutput(), Pose3f());
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/PoseDriftCompensator.cpp b/media/libheadtracking/PoseDriftCompensator.cpp
new file mode 100644
index 0000000..0e90cad
--- /dev/null
+++ b/media/libheadtracking/PoseDriftCompensator.cpp
@@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"){}
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "PoseDriftCompensator.h"
+
+#include <cmath>
+
+#include "QuaternionUtil.h"
+
+namespace android {
+namespace media {
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+PoseDriftCompensator::PoseDriftCompensator(const Options& options) : mOptions(options) {}
+
+void PoseDriftCompensator::setInput(int64_t timestamp, const Pose3f& input) {
+    if (mTimestamp.has_value()) {
+        // Avoid computation upon first input (only sets the initial state).
+        Pose3f prevInputToInput = mPrevInput.inverse() * input;
+        mOutput = scale(mOutput, timestamp - mTimestamp.value()) * prevInputToInput;
+    }
+    mPrevInput = input;
+    mTimestamp = timestamp;
+}
+
+void PoseDriftCompensator::recenter() {
+    mTimestamp.reset();
+    mOutput = Pose3f();
+}
+
+Pose3f PoseDriftCompensator::getOutput() const {
+    return mOutput;
+}
+
+Pose3f PoseDriftCompensator::scale(const Pose3f& pose, int64_t dt) {
+    // Translation.
+    Vector3f translation = pose.translation();
+    translation *= std::expf(-static_cast<float>(dt) / mOptions.translationalDriftTimeConstant);
+
+    // Rotation.
+    Vector3f rotationVec = quaternionToRotationVector(pose.rotation());
+    rotationVec *= std::expf(-static_cast<float>(dt) / mOptions.rotationalDriftTimeConstant);
+
+    return Pose3f(translation, rotationVectorToQuaternion(rotationVec));
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/PoseDriftCompensator.h b/media/libheadtracking/PoseDriftCompensator.h
new file mode 100644
index 0000000..a71483b
--- /dev/null
+++ b/media/libheadtracking/PoseDriftCompensator.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <optional>
+
+#include "media/Pose.h"
+
+namespace android {
+namespace media {
+
+/**
+ * Drift compensator for a stream of poses.
+ *
+ * This is effectively a high-pass filter for a pose stream, removing any DC-offset / bias. The
+ * provided input stream will be "pulled" toward identity with an exponential decay filter with a
+ * configurable time constant. Rotation and translation are handled separately.
+ *
+ * Typical usage:
+ * PoseDriftCompensator comp(...);
+ *
+ * while (...) {
+ *   comp.setInput(...);
+ *   Pose3f output = comp.getOutput();
+ * }
+ *
+ * There doesn't need to be a 1:1 correspondence between setInput() and getOutput() calls. The
+ * output timestamp is always that of the last setInput() call. Calling recenter() will reset the
+ * bias to the current output, causing the output to be identity.
+ *
+ * The initial bias point is identity.
+ *
+ * This implementation is thread-compatible, but not thread-safe.
+ */
+class PoseDriftCompensator {
+  public:
+    struct Options {
+        float translationalDriftTimeConstant = std::numeric_limits<float>::infinity();
+        float rotationalDriftTimeConstant = std::numeric_limits<float>::infinity();
+    };
+
+    explicit PoseDriftCompensator(const Options& options);
+
+    void setInput(int64_t timestamp, const Pose3f& input);
+
+    void recenter();
+
+    Pose3f getOutput() const;
+
+  private:
+    const Options mOptions;
+
+    Pose3f mPrevInput;
+    Pose3f mOutput;
+    std::optional<int64_t> mTimestamp;
+
+    Pose3f scale(const Pose3f& pose, int64_t dt);
+};
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/PoseProcessingGraph.png b/media/libheadtracking/PoseProcessingGraph.png
new file mode 100644
index 0000000..0363068
--- /dev/null
+++ b/media/libheadtracking/PoseProcessingGraph.png
Binary files differ
diff --git a/media/libheadtracking/PoseRateLimiter-test.cpp b/media/libheadtracking/PoseRateLimiter-test.cpp
new file mode 100644
index 0000000..f306183
--- /dev/null
+++ b/media/libheadtracking/PoseRateLimiter-test.cpp
@@ -0,0 +1,101 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <gtest/gtest.h>
+
+#include "PoseRateLimiter.h"
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+namespace android {
+namespace media {
+namespace {
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+using Options = PoseRateLimiter::Options;
+
+TEST(PoseRateLimiter, Initial) {
+    Pose3f target({1, 2, 3}, Quaternionf::UnitRandom());
+    PoseRateLimiter limiter(Options{.maxTranslationalVelocity = 10, .maxRotationalVelocity = 10});
+    limiter.setTarget(target);
+    EXPECT_EQ(limiter.calculatePose(1000), target);
+}
+
+TEST(PoseRateLimiter, UnlimitedZeroTime) {
+    Pose3f target1({1, 2, 3}, Quaternionf::UnitRandom());
+    Pose3f target2({4, 5, 6}, Quaternionf::UnitRandom());
+    PoseRateLimiter limiter(Options{});
+    limiter.setTarget(target1);
+    EXPECT_EQ(limiter.calculatePose(0), target1);
+    limiter.setTarget(target2);
+    EXPECT_EQ(limiter.calculatePose(0), target2);
+    limiter.setTarget(target1);
+    EXPECT_EQ(limiter.calculatePose(0), target1);
+}
+
+TEST(PoseRateLimiter, Limited) {
+    Pose3f pose1({1, 2, 3}, Quaternionf::Identity());
+    Pose3f pose2({1, 2, 8}, rotateZ(M_PI * 5 / 8));
+    PoseRateLimiter limiter(Options{.maxTranslationalVelocity = 1, .maxRotationalVelocity = 10});
+    limiter.setTarget(pose2);
+    EXPECT_EQ(limiter.calculatePose(1000), pose2);
+
+    // Rate limiting is inactive. Should track despite the violation.
+    limiter.setTarget(pose1);
+    EXPECT_EQ(limiter.calculatePose(1001), pose1);
+
+    // Enable rate limiting and observe gradual motion from pose1 to pose2.
+    limiter.enable();
+    limiter.setTarget(pose2);
+    EXPECT_EQ(limiter.calculatePose(1002), Pose3f({1, 2, 4}, rotateZ(M_PI * 1 / 8)));
+    limiter.setTarget(pose2);
+    EXPECT_EQ(limiter.calculatePose(1003), Pose3f({1, 2, 5}, rotateZ(M_PI * 2 / 8)));
+    // Skip a tick.
+    limiter.setTarget(pose2);
+    EXPECT_EQ(limiter.calculatePose(1005), Pose3f({1, 2, 7}, rotateZ(M_PI * 4 / 8)));
+    limiter.setTarget(pose2);
+    EXPECT_EQ(limiter.calculatePose(1006), pose2);
+
+    // We reached the target, so rate limiting should now be disabled.
+    limiter.setTarget(pose1);
+    EXPECT_EQ(limiter.calculatePose(1007), pose1);
+}
+
+TEST(PoseRateLimiter, Reset) {
+    Pose3f pose1({1, 2, 3}, Quaternionf::Identity());
+    Pose3f pose2({1, 2, 8}, rotateZ(M_PI * 5 / 8));
+    PoseRateLimiter limiter(Options{.maxTranslationalVelocity = 1, .maxRotationalVelocity = 10});
+    limiter.setTarget(pose1);
+    EXPECT_EQ(limiter.calculatePose(1000), pose1);
+
+    // Enable rate limiting and observe gradual motion from pose1 to pose2.
+    limiter.enable();
+    limiter.setTarget(pose2);
+    EXPECT_EQ(limiter.calculatePose(1001), Pose3f({1, 2, 4}, rotateZ(M_PI * 1 / 8)));
+
+    // Reset the pose and disable rate limiting.
+    limiter.reset(pose2);
+    EXPECT_EQ(limiter.calculatePose(1002), pose2);
+
+    // Rate limiting should now be disabled.
+    limiter.setTarget(pose1);
+    EXPECT_EQ(limiter.calculatePose(1003), pose1);
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/PoseRateLimiter.cpp b/media/libheadtracking/PoseRateLimiter.cpp
new file mode 100644
index 0000000..380e22b
--- /dev/null
+++ b/media/libheadtracking/PoseRateLimiter.cpp
@@ -0,0 +1,52 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"){}
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "PoseRateLimiter.h"
+
+namespace android {
+namespace media {
+
+PoseRateLimiter::PoseRateLimiter(const Options& options) : mOptions(options), mLimiting(false) {}
+
+void PoseRateLimiter::enable() {
+    mLimiting = true;
+}
+
+void PoseRateLimiter::reset(const Pose3f& target) {
+    mLimiting = false;
+    mTargetPose = target;
+}
+
+void PoseRateLimiter::setTarget(const Pose3f& target) {
+    mTargetPose = target;
+}
+
+Pose3f PoseRateLimiter::calculatePose(int64_t timestamp) {
+    assert(mTargetPose.has_value());
+    Pose3f pose;
+    if (mLimiting && mOutput.has_value()) {
+        std::tie(pose, mLimiting) = moveWithRateLimit(
+                mOutput->pose, mTargetPose.value(), timestamp - mOutput->timestamp,
+                mOptions.maxTranslationalVelocity, mOptions.maxRotationalVelocity);
+    } else {
+        pose = mTargetPose.value();
+    }
+    mOutput = Point{pose, timestamp};
+    return pose;
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/PoseRateLimiter.h b/media/libheadtracking/PoseRateLimiter.h
new file mode 100644
index 0000000..aa2fe80
--- /dev/null
+++ b/media/libheadtracking/PoseRateLimiter.h
@@ -0,0 +1,93 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <optional>
+
+#include "media/Pose.h"
+
+namespace android {
+namespace media {
+
+/**
+ * Limits a stream of poses to a given maximum translational and rotational velocities.
+ *
+ * Normal operation:
+ *
+ * Pose3f output;
+ * PoseRateLimiter limiter(...);
+ *
+ * // Limiting is disabled. Output will be the same as last input.
+ * limiter.setTarget(...);
+ * output = limiter.calculatePose(...);
+ * limiter.setTarget(...);
+ * output = limiter.calculatePose(...);
+ *
+ * // Enable limiting. Output will no longer be necessarily the same as last input.
+ * limiter.enable();
+ * limiter.setTarget(...);
+ * output = limiter.calculatePose(...);
+ * limiter.setTarget(...);
+ * output = limiter.calculatePose(...);
+ *
+ * // When eventually the output has been able to catch up with the last input, the limited will be
+ * // automatically disabled again and the output will match the input again.
+ * limiter.setTarget(...);
+ * output = limiter.calculatePose(...);
+ *
+ * As shown above, the limiter is turned on manually via enable(), but turns off automatically as
+ * soon as the output is able to catch up to the input. The intention is that rate limiting will be
+ * turned on at specific times to smooth out any artificial discontinuities introduced to the pose
+ * stream, but the rest of the time will be a simple passthrough.
+
+ * setTarget(...) and calculatePose(...) don't have to be ordered in any particular way. However,
+ * setTarget or reset() must be called at least once prior to the first calculatePose().
+ *
+ * Calling reset() instead of setTarget() forces the output to the given pose and disables rate
+ * limiting.
+ *
+ * This implementation is thread-compatible, but not thread-safe.
+ */
+class PoseRateLimiter {
+  public:
+    struct Options {
+        float maxTranslationalVelocity = std::numeric_limits<float>::infinity();
+        float maxRotationalVelocity = std::numeric_limits<float>::infinity();
+    };
+
+    explicit PoseRateLimiter(const Options& options);
+
+    void enable();
+
+    void reset(const Pose3f& target);
+    void setTarget(const Pose3f& target);
+
+    Pose3f calculatePose(int64_t timestamp);
+
+  private:
+    struct Point {
+        Pose3f pose;
+        int64_t timestamp;
+    };
+
+    const Options mOptions;
+    bool mLimiting;
+    std::optional<Pose3f> mTargetPose;
+    std::optional<Point> mOutput;
+};
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/QuaternionUtil-test.cpp b/media/libheadtracking/QuaternionUtil-test.cpp
new file mode 100644
index 0000000..e79e54a
--- /dev/null
+++ b/media/libheadtracking/QuaternionUtil-test.cpp
@@ -0,0 +1,56 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <gtest/gtest.h>
+
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+namespace android {
+namespace media {
+namespace {
+
+TEST(QuaternionUtil, RotationVectorToQuaternion) {
+    // 90 degrees around Z.
+    Vector3f rot = {0, 0, M_PI_2};
+    Quaternionf quat = rotationVectorToQuaternion(rot);
+    ASSERT_EQ(quat * Vector3f(1, 0, 0), Vector3f(0, 1, 0));
+    ASSERT_EQ(quat * Vector3f(0, 1, 0), Vector3f(-1, 0, 0));
+    ASSERT_EQ(quat * Vector3f(0, 0, 1), Vector3f(0, 0, 1));
+}
+
+TEST(QuaternionUtil, QuaternionToRotationVector) {
+    Quaternionf quat = Quaternionf::FromTwoVectors(Vector3f(1, 0, 0), Vector3f(0, 1, 0));
+    Vector3f rot = quaternionToRotationVector(quat);
+    ASSERT_EQ(rot, Vector3f(0, 0, M_PI_2));
+}
+
+TEST(QuaternionUtil, RoundTripFromQuaternion) {
+    Quaternionf quaternion = Quaternionf::UnitRandom();
+    EXPECT_EQ(quaternion, rotationVectorToQuaternion(quaternionToRotationVector(quaternion)));
+}
+
+TEST(QuaternionUtil, RoundTripFromVector) {
+    Vector3f vec{0.1, 0.2, 0.3};
+    EXPECT_EQ(vec, quaternionToRotationVector(rotationVectorToQuaternion(vec)));
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/QuaternionUtil.cpp b/media/libheadtracking/QuaternionUtil.cpp
new file mode 100644
index 0000000..5d090de
--- /dev/null
+++ b/media/libheadtracking/QuaternionUtil.cpp
@@ -0,0 +1,103 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "QuaternionUtil.h"
+
+#include <cassert>
+
+namespace android {
+namespace media {
+
+using Eigen::NumTraits;
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+namespace {
+
+Vector3f LogSU2(const Quaternionf& q) {
+    // Implementation of the logarithmic map of SU(2) using atan.
+    // This follows Hertzberg et al. "Integrating Generic Sensor Fusion Algorithms
+    // with Sound State Representations through Encapsulation of Manifolds", Eq.
+    // (31)
+    // We use asin and acos instead of atan to enable the use of Eigen Autodiff
+    // with SU2.
+    const float sign_of_w = q.w() < 0.f ? -1.f : 1.f;
+    const float abs_w = sign_of_w * q.w();
+    const Vector3f v = sign_of_w * q.vec();
+    const float squared_norm_of_v = v.squaredNorm();
+
+    assert(abs(1.f - abs_w * abs_w - squared_norm_of_v) < NumTraits<float>::dummy_precision());
+
+    if (squared_norm_of_v > NumTraits<float>::dummy_precision()) {
+        const float norm_of_v = sqrt(squared_norm_of_v);
+        if (abs_w > NumTraits<float>::dummy_precision()) {
+            // asin(x) = acos(x) at x = 1/sqrt(2).
+            if (norm_of_v <= float(M_SQRT1_2)) {
+                return (asin(norm_of_v) / norm_of_v) * v;
+            }
+            return (acos(abs_w) / norm_of_v) * v;
+        }
+        return (M_PI_2 / norm_of_v) * v;
+    }
+
+    // Taylor expansion at squared_norm_of_v == 0
+    return (1.f / abs_w - squared_norm_of_v / (3.f * pow(abs_w, 3))) * v;
+}
+
+Quaternionf ExpSU2(const Vector3f& delta) {
+    Quaternionf q_delta;
+    const float theta_squared = delta.squaredNorm();
+    if (theta_squared > NumTraits<float>::dummy_precision()) {
+        const float theta = sqrt(theta_squared);
+        q_delta.w() = cos(theta);
+        q_delta.vec() = (sin(theta) / theta) * delta;
+    } else {
+        // taylor expansions around theta == 0
+        q_delta.w() = 1.f - 0.5f * theta_squared;
+        q_delta.vec() = (1.f - 1.f / 6.f * theta_squared) * delta;
+    }
+    return q_delta;
+}
+
+}  // namespace
+
+Quaternionf rotationVectorToQuaternion(const Vector3f& rotationVector) {
+    //  SU(2) is a double cover of SO(3), thus we have to half the tangent vector
+    //  delta
+    const Vector3f half_delta = 0.5f * rotationVector;
+    return ExpSU2(half_delta);
+}
+
+Vector3f quaternionToRotationVector(const Quaternionf& quaternion) {
+    // SU(2) is a double cover of SO(3), thus we have to multiply the tangent
+    // vector delta by two
+    return 2.f * LogSU2(quaternion);
+}
+
+Quaternionf rotateX(float angle) {
+    return rotationVectorToQuaternion(Vector3f(1, 0, 0) * angle);
+}
+
+Quaternionf rotateY(float angle) {
+    return rotationVectorToQuaternion(Vector3f(0, 1, 0) * angle);
+}
+
+Quaternionf rotateZ(float angle) {
+    return rotationVectorToQuaternion(Vector3f(0, 0, 1) * angle);
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/QuaternionUtil.h b/media/libheadtracking/QuaternionUtil.h
new file mode 100644
index 0000000..f7a2ca9
--- /dev/null
+++ b/media/libheadtracking/QuaternionUtil.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <Eigen/Geometry>
+
+namespace android {
+namespace media {
+
+/**
+ * Converts a rotation vector to an equivalent quaternion.
+ * The rotation vector is given as a 3-vector whose direction represents the rotation axis and its
+ * magnitude the rotation angle (in radians) around that axis.
+ */
+Eigen::Quaternionf rotationVectorToQuaternion(const Eigen::Vector3f& rotationVector);
+
+/**
+ * Converts a quaternion to an equivalent rotation vector.
+ * The rotation vector is given as a 3-vector whose direction represents the rotation axis and its
+ * magnitude the rotation angle (in radians) around that axis.
+ */
+Eigen::Vector3f quaternionToRotationVector(const Eigen::Quaternionf& quaternion);
+
+/**
+ * Returns a quaternion representing a rotation around the X-axis with the given amount (in
+ * radians).
+ */
+Eigen::Quaternionf rotateX(float angle);
+
+/**
+ * Returns a quaternion representing a rotation around the Y-axis with the given amount (in
+ * radians).
+ */
+Eigen::Quaternionf rotateY(float angle);
+
+/**
+ * Returns a quaternion representing a rotation around the Z-axis with the given amount (in
+ * radians).
+ */
+Eigen::Quaternionf rotateZ(float angle);
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/README.md b/media/libheadtracking/README.md
new file mode 100644
index 0000000..3d5b71a
--- /dev/null
+++ b/media/libheadtracking/README.md
@@ -0,0 +1,185 @@
+# Head-Tracking Library For Immersive Audio
+
+This library handles the processing of head-tracking information, necessary for
+Immersive Audio functionality. It goes from bare sensor reading into the final
+pose fed into a virtualizer.
+
+## Basic Usage
+
+The main entry point into this library is the `HeadTrackingProcessor` class.
+This class is provided with the following inputs:
+
+- Head pose, relative to some arbitrary world frame.
+- Screen pose, relative to some arbitrary world frame.
+- Display orientation, defined as the angle between the "physical" screen and
+  the "logical" screen.
+- Transform between the screen and the sound stage.
+- Desired operational mode:
+    - Static: only the sound stage pose is taken into account. This will result
+      in an experience where the sound stage moved with the listener's head.
+    - World-relative: both the head pose and stage pose are taken into account.
+      This will result in an experience where the sound stage is perceived to be
+      located at a fixed place in the world.
+    - Screen-relative: the head pose, screen pose and stage pose are all taken
+      into account. This will result in an experience where the sound stage is
+      perceived to be located at a fixed place relative to the screen.
+
+Once inputs are provided, the `calculate()` method will make the following
+output available:
+
+- Stage pose, relative to the head. This aggregates all the inputs mentioned
+  above and is ready to be fed into a virtualizer.
+- Actual operational mode. May deviate from the desired one in cases where the
+  desired mode cannot be calculated (for example, as result of dropped messages
+  from one of the sensors).
+
+A `recenter()` operation is also available, which indicates to the system that
+whatever pose the screen and head are currently at should be considered as the
+"center" pose, or frame of reference.
+
+## Pose-Related Conventions
+
+### Naming and Composition
+
+When referring to poses in code, it is always good practice to follow
+conventional naming, which highlights the reference and target frames clearly:
+
+Bad:
+
+```
+Pose3f headPose;
+```
+
+Good:
+
+```
+Pose3f worldToHead;  // “world” is the reference frame,
+                     // “head” is the target frame.
+```
+
+By following this convention, it is easy to follow correct composition of poses,
+by making sure adjacent frames are identical:
+
+```
+Pose3f aToD = aToB * bToC * cToD;
+```
+
+And similarly, inverting the transform simply flips the reference and target:
+
+```
+Pose3f aToB = bToA.inverse();
+```
+
+### Twist
+
+“Twist” is to pose what velocity is to distance: it is the time-derivative of a
+pose, representing the change in pose over a short period of time. Its naming
+convention always states one frame, e.g.:
+Twist3f headTwist;
+
+This means that this twist represents the head-at-time-T to head-at-time-T+dt
+transform. Twists are not composable in the same way as poses.
+
+### Frames of Interest
+
+The frames of interest in this library are defined as follows:
+
+#### Head
+
+This is the listener’s head. The origin is at the center point between the
+ear-drums, the X-axis goes from left ear to right ear, Y-axis goes from the back
+of the head towards the face and Z-axis goes from the bottom of the head to the
+top.
+
+#### Screen
+
+This is the primary screen that the user will be looking at, which is relevant
+for some Immersive Audio use-cases, such as watching a movie. We will follow a
+different convention for this frame than what the Sensor framework uses. The
+origin is at the center of the screen. X-axis goes from left to right, Z-axis
+goes from the screen bottom to the screen top, Y-axis goes “into” the screen (
+from the direction of the viewer). The up/down/left/right of the screen are
+defined as the logical directions used for display. So when flipping the display
+orientation between “landscape” and “portrait”, the frame of reference will
+change with respect to the physical screen.
+
+#### Stage
+
+This is the frame of reference used by the virtualizer for positioning sound
+objects. It is not associated with any physical frame. In a typical
+multi-channel scenario, the listener is at the origin, the X-axis goes from left
+to right, Y-axis from back to front and Z-axis from down to up. For example, a
+front-right speaker is located at positive X, Y and Z=0, a height speaker will
+have a positive Z.
+
+#### World
+
+It is sometimes convenient to use an intermediate frame when dealing with
+head-to-screen transforms. The “world” frame is an arbitrary frame of reference
+in the physical world, relative to which we can measure the head pose and screen
+pose. In (very common) cases when we can’t establish such an absolute frame, we
+can take each measurement relative to a separate, arbitrary frame and high-pass
+the result.
+
+## Processing Description
+
+![Pose processing graph](PoseProcessingGraph.png)
+
+The diagram above illustrates the processing that takes place from the inputs to
+the outputs.
+
+### Predictor
+
+The Predictor block gets pose + twist (pose derivative) and extrapolates to
+obtain a predicted head pose (w/ given latency).
+
+### Drift / Bias Compensator
+
+The Drift / Bias Compensator blocks serve two purposes:
+
+- Compensate for floating reference axes by applying a high-pass filter, which
+  slowly pulls the pose toward identity.
+- Establish the reference frame for the poses by having the ability to set the
+  current pose as the reference for future poses (recentering). Effectively,
+  this is resetting the filter state to identity.
+
+### Orientation Compensation
+
+The Orientation Compensation block applies the display orientation to the screen
+pose to obtain the pose of the “logical screen” frame, in which the Y-axis is
+pointing in the direction of the logical screen “up” rather than the physical
+one.
+
+### Screen-Relative Pose
+
+The Screen-Relative Pose block is provided with a head pose and a screen pose
+and estimates the pose of the head relative to the screen. Optionally, this
+module may indicate that the user is likely not in front of the screen via the
+“valid” output.
+
+### Mode Selector
+
+The Mode Selector block aggregates the various sources of pose information into
+a head-to-stage pose that is going to feed the virtualizer. It is controlled by
+the “desired mode” signal that indicates whether the preference is to be in
+either static, world-relative or screen-relative.
+
+The actual mode may diverge from the desired mode. It is determined as follows:
+
+- If the desired mode is static, the actual mode is static.
+- If the desired mode is world-relative:
+    - If head poses are fresh, the actual mode is world-relative.
+    - Otherwise the actual mode is static.
+- If the desired mode is screen-relative:
+    - If head and screen poses are fresh and the ‘valid’ signal is asserted, the
+      actual mode is screen-relative.
+    - Otherwise, apply the same rules as the desired mode being world-relative.
+
+### Rate Limiter
+
+A Rate Limiter block is applied to the final output to smooth out any abrupt
+transitions caused by any of the following events:
+
+- Mode switch.
+- Display orientation switch.
+- Recenter operation.
diff --git a/media/libheadtracking/ScreenHeadFusion-test.cpp b/media/libheadtracking/ScreenHeadFusion-test.cpp
new file mode 100644
index 0000000..ecf27f5
--- /dev/null
+++ b/media/libheadtracking/ScreenHeadFusion-test.cpp
@@ -0,0 +1,75 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <gtest/gtest.h>
+
+#include "ScreenHeadFusion.h"
+#include "TestUtil.h"
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+namespace android {
+namespace media {
+namespace {
+
+TEST(ScreenHeadFusion, Init) {
+    ScreenHeadFusion fusion;
+    EXPECT_FALSE(fusion.calculate().has_value());
+}
+
+TEST(ScreenHeadFusion, Calculate_NoHead) {
+    ScreenHeadFusion fusion;
+    fusion.setWorldToScreenPose(0, Pose3f());
+    EXPECT_FALSE(fusion.calculate().has_value());
+}
+
+TEST(ScreenHeadFusion, Calculate_NoScreen) {
+    ScreenHeadFusion fusion;
+    fusion.setWorldToHeadPose(0, Pose3f());
+    EXPECT_FALSE(fusion.calculate().has_value());
+}
+
+TEST(ScreenHeadFusion, Calculate) {
+    Pose3f worldToScreen1({1, 2, 3}, Quaternionf::UnitRandom());
+    Pose3f worldToHead1({4, 5, 6}, Quaternionf::UnitRandom());
+    Pose3f worldToScreen2({11, 12, 13}, Quaternionf::UnitRandom());
+    Pose3f worldToHead2({14, 15, 16}, Quaternionf::UnitRandom());
+
+    ScreenHeadFusion fusion;
+    fusion.setWorldToHeadPose(123, worldToHead1);
+    fusion.setWorldToScreenPose(456, worldToScreen1);
+    auto result = fusion.calculate();
+    ASSERT_TRUE(result.has_value());
+    EXPECT_EQ(123, result->timestamp);
+    EXPECT_EQ(worldToScreen1.inverse() * worldToHead1, result->pose);
+
+    fusion.setWorldToHeadPose(567, worldToHead2);
+    result = fusion.calculate();
+    ASSERT_TRUE(result.has_value());
+    EXPECT_EQ(456, result->timestamp);
+    EXPECT_EQ(worldToScreen1.inverse() * worldToHead2, result->pose);
+
+    fusion.setWorldToScreenPose(678, worldToScreen2);
+    result = fusion.calculate();
+    ASSERT_TRUE(result.has_value());
+    EXPECT_EQ(567, result->timestamp);
+    EXPECT_EQ(worldToScreen2.inverse() * worldToHead2, result->pose);
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/ScreenHeadFusion.cpp b/media/libheadtracking/ScreenHeadFusion.cpp
new file mode 100644
index 0000000..f023570
--- /dev/null
+++ b/media/libheadtracking/ScreenHeadFusion.cpp
@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"){}
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "ScreenHeadFusion.h"
+
+namespace android {
+namespace media {
+
+void ScreenHeadFusion::setWorldToHeadPose(int64_t timestamp, const Pose3f& worldToHead) {
+    mWorldToHead = TimestampedPose{.timestamp = timestamp, .pose = worldToHead};
+}
+
+void ScreenHeadFusion::setWorldToScreenPose(int64_t timestamp, const Pose3f& worldToScreen) {
+    mWorldToScreen = TimestampedPose{.timestamp = timestamp, .pose = worldToScreen};
+}
+
+std::optional<ScreenHeadFusion::TimestampedPose> ScreenHeadFusion::calculate() {
+    // TODO: this is temporary, simplistic logic.
+    if (!mWorldToHead.has_value() || !mWorldToScreen.has_value()) {
+        return std::nullopt;
+    }
+    return TimestampedPose{
+            .timestamp = std::min(mWorldToHead->timestamp, mWorldToScreen->timestamp),
+            .pose = mWorldToScreen->pose.inverse() * mWorldToHead->pose};
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/ScreenHeadFusion.h b/media/libheadtracking/ScreenHeadFusion.h
new file mode 100644
index 0000000..ee81100
--- /dev/null
+++ b/media/libheadtracking/ScreenHeadFusion.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <optional>
+
+#include "media/Pose.h"
+
+namespace android {
+namespace media {
+
+/**
+ * Combines world-to-head pose with world-to-screen pose to obtain screen-to-head.
+ *
+ * Input poses may arrive separately. The last pose of each kind is taken into account. The
+ * timestamp of the output is the ealier (older) timestamp of the two inputs.
+ *
+ * Output may be nullopt in the following cases:
+ * - Either one of the inputs has not yet been provided.
+ * - It is estimated that the user is no longer facing the screen.
+ *
+ * Typical usage:
+ *
+ * ScreenHeadFusion fusion(...);
+ * fusion.setWorldToHeadPose(...);
+ * fusion.setWorldToScreenPose(...);
+ * auto output = fusion.calculate();
+ *
+ * This class is not thread-safe, but thread-compatible.
+ */
+class ScreenHeadFusion {
+  public:
+    struct TimestampedPose {
+        int64_t timestamp;
+        Pose3f pose;
+    };
+
+    void setWorldToHeadPose(int64_t timestamp, const Pose3f& worldToHead);
+
+    void setWorldToScreenPose(int64_t timestamp, const Pose3f& worldToScreen);
+
+    /**
+     * Returns the screen-to-head pose, or nullopt if invalid.
+     */
+    std::optional<TimestampedPose> calculate();
+
+  private:
+    std::optional<TimestampedPose> mWorldToHead;
+    std::optional<TimestampedPose> mWorldToScreen;
+};
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/SensorPoseProvider-example.cpp b/media/libheadtracking/SensorPoseProvider-example.cpp
new file mode 100644
index 0000000..13db07e
--- /dev/null
+++ b/media/libheadtracking/SensorPoseProvider-example.cpp
@@ -0,0 +1,76 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <unistd.h>
+#include <iostream>
+
+#include <android/sensor.h>
+#include <hardware/sensors.h>
+#include <utils/SystemClock.h>
+
+#include <media/SensorPoseProvider.h>
+
+using android::elapsedRealtimeNano;
+using android::media::Pose3f;
+using android::media::SensorPoseProvider;
+using android::media::Twist3f;
+
+const char kPackageName[] = "SensorPoseProvider-example";
+
+class Listener : public SensorPoseProvider::Listener {
+  public:
+    void onPose(int64_t timestamp, int32_t handle, const Pose3f& pose,
+                const std::optional<Twist3f>& twist) override {
+        int64_t now = elapsedRealtimeNano();
+
+        std::cout << "onPose t=" << timestamp
+                  << " lag=" << ((now - timestamp) / 1e6) << "[ms]"
+                  << " sensor=" << handle
+                  << " pose=" << pose
+                  << " twist=";
+        if (twist.has_value()) {
+            std::cout << twist.value();
+        } else {
+            std::cout << "<none>";
+        }
+        std::cout << std::endl;
+    }
+};
+
+int main() {
+    ASensorManager* sensor_manager = ASensorManager_getInstanceForPackage(kPackageName);
+    if (!sensor_manager) {
+        std::cerr << "Failed to get a sensor manager" << std::endl;
+        return 1;
+    }
+
+    const ASensor* headSensor =
+            ASensorManager_getDefaultSensor(sensor_manager, SENSOR_TYPE_GAME_ROTATION_VECTOR);
+    const ASensor* screenSensor =
+            ASensorManager_getDefaultSensor(sensor_manager, SENSOR_TYPE_ROTATION_VECTOR);
+
+    Listener listener;
+
+    std::unique_ptr<SensorPoseProvider> provider =
+            SensorPoseProvider::create(kPackageName, &listener);
+    int32_t headHandle = provider->startSensor(headSensor, std::chrono::milliseconds(500));
+    sleep(2);
+    provider->startSensor(screenSensor, std::chrono::milliseconds(500));
+    sleep(2);
+    provider->stopSensor(headHandle);
+    sleep(2);
+    return 0;
+}
diff --git a/media/libheadtracking/SensorPoseProvider.cpp b/media/libheadtracking/SensorPoseProvider.cpp
new file mode 100644
index 0000000..cfa23d0
--- /dev/null
+++ b/media/libheadtracking/SensorPoseProvider.cpp
@@ -0,0 +1,241 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <media/SensorPoseProvider.h>
+
+#define LOG_TAG "SensorPoseProvider"
+
+#include <inttypes.h>
+
+#include <future>
+#include <map>
+#include <thread>
+
+#include <android/looper.h>
+#include <log/log_main.h>
+
+#include "QuaternionUtil.h"
+
+namespace android {
+namespace media {
+namespace {
+
+/**
+ * RAII-wrapper around ASensorEventQueue, which destroys it on destruction.
+ */
+class EventQueueGuard {
+  public:
+    EventQueueGuard(ASensorManager* manager, ASensorEventQueue* queue)
+        : mManager(manager), mQueue(queue) {}
+
+    ~EventQueueGuard() {
+        if (mQueue) {
+            int ret = ASensorManager_destroyEventQueue(mManager, mQueue);
+            if (ret) {
+                ALOGE("Failed to destroy event queue: %s\n", strerror(ret));
+            }
+        }
+    }
+
+    EventQueueGuard(const EventQueueGuard&) = delete;
+    EventQueueGuard& operator=(const EventQueueGuard&) = delete;
+
+    [[nodiscard]] ASensorEventQueue* get() const { return mQueue; }
+
+  private:
+    ASensorManager* const mManager;
+    ASensorEventQueue* mQueue;
+};
+
+/**
+ * RAII-wrapper around an enabled sensor, which disables it upon destruction.
+ */
+class SensorEnableGuard {
+  public:
+    SensorEnableGuard(ASensorEventQueue* queue, const ASensor* sensor)
+        : mQueue(queue), mSensor(sensor) {}
+
+    ~SensorEnableGuard() {
+        if (mSensor) {
+            int ret = ASensorEventQueue_disableSensor(mQueue, mSensor);
+            if (ret) {
+                ALOGE("Failed to disable sensor: %s\n", strerror(ret));
+            }
+        }
+    }
+
+    SensorEnableGuard(const SensorEnableGuard&) = delete;
+    SensorEnableGuard& operator=(const SensorEnableGuard&) = delete;
+
+    // Enable moving.
+    SensorEnableGuard(SensorEnableGuard&& other) : mQueue(other.mQueue), mSensor(other.mSensor) {
+        other.mSensor = nullptr;
+    }
+
+  private:
+    ASensorEventQueue* const mQueue;
+    const ASensor* mSensor;
+};
+
+/**
+ * Streams the required events to a PoseListener, based on events originating from the Sensor stack.
+ */
+class SensorPoseProviderImpl : public SensorPoseProvider {
+  public:
+    static std::unique_ptr<SensorPoseProvider> create(const char* packageName, Listener* listener) {
+        std::unique_ptr<SensorPoseProviderImpl> result(
+                new SensorPoseProviderImpl(packageName, listener));
+        return result->waitInitFinished() ? std::move(result) : nullptr;
+    }
+
+    ~SensorPoseProviderImpl() override {
+        // Disable all active sensors.
+        mEnabledSensors.clear();
+        ALooper_wake(mLooper);
+        mThread.join();
+    }
+
+    int32_t startSensor(const ASensor* sensor, std::chrono::microseconds samplingPeriod) override {
+        int32_t handle = ASensor_getHandle(sensor);
+
+        // Enable the sensor.
+        if (ASensorEventQueue_registerSensor(mQueue, sensor, samplingPeriod.count(), 0)) {
+            ALOGE("Failed to enable sensor");
+            return INVALID_HANDLE;
+        }
+
+        mEnabledSensors.emplace(handle, SensorEnableGuard(mQueue, sensor));
+        return handle;
+    }
+
+    void stopSensor(int handle) override { mEnabledSensors.erase(handle); }
+
+  private:
+    ALooper* mLooper;
+    Listener* const mListener;
+
+    std::thread mThread;
+    std::map<int32_t, SensorEnableGuard> mEnabledSensors;
+    ASensorEventQueue* mQueue;
+
+    // We must do some of the initialization operations on the worker thread, because the API relies
+    // on the thread-local looper. In addition, as a matter of convenience, we store some of the
+    // state on the stack.
+    // For that reason, we use a two-step initialization approach, where the ctor mostly just starts
+    // the worker thread and that thread would notify, via the promise below whenever initialization
+    // is finished, and whether it was successful.
+    std::promise<bool> mInitPromise;
+
+    SensorPoseProviderImpl(const char* packageName, Listener* listener)
+        : mListener(listener),
+          mThread([this, p = std::string(packageName)] { threadFunc(p.c_str()); }) {}
+
+    void initFinished(bool success) { mInitPromise.set_value(success); }
+
+    bool waitInitFinished() { return mInitPromise.get_future().get(); }
+
+    void threadFunc(const char* packageName) {
+        // Obtain looper.
+        mLooper = ALooper_prepare(ALOOPER_PREPARE_ALLOW_NON_CALLBACKS);
+
+        // The number 19 is arbitrary, only useful if using multiple objects on the same looper.
+        constexpr int kIdent = 19;
+
+        // Obtain sensor manager.
+        ASensorManager* sensor_manager = ASensorManager_getInstanceForPackage(packageName);
+        if (!sensor_manager) {
+            ALOGE("Failed to get a sensor manager");
+            initFinished(false);
+            return;
+        }
+
+        // Create event queue.
+        mQueue = ASensorManager_createEventQueue(sensor_manager, mLooper, kIdent, nullptr, nullptr);
+
+        if (mQueue == nullptr) {
+            ALOGE("Failed to create a sensor event queue");
+            initFinished(false);
+            return;
+        }
+
+        EventQueueGuard eventQueueGuard(sensor_manager, mQueue);
+
+        initFinished(true);
+
+        while (true) {
+            int ret = ALooper_pollOnce(-1 /* no timeout */, nullptr, nullptr, nullptr);
+
+            switch (ret) {
+                case ALOOPER_POLL_WAKE:
+                    // Normal way to exit.
+                    return;
+
+                case kIdent:
+                    // Possible events on our queue.
+                    break;
+
+                default:
+                    ALOGE("Unexpected status out of ALooper_pollOnce: %d", ret);
+            }
+
+            // Process an event.
+            ASensorEvent event;
+            ssize_t size = ASensorEventQueue_getEvents(mQueue, &event, 1);
+            if (size < 0 || size > 1) {
+                ALOGE("Unexpected return value from ASensorEventQueue_getEvents: %zd", size);
+                break;
+            }
+            if (size == 0) {
+                // No events.
+                continue;
+            }
+
+            handleEvent(event);
+        }
+    }
+
+    void handleEvent(const ASensorEvent& event) {
+        auto value = parseEvent(event);
+        mListener->onPose(event.timestamp, event.sensor, std::get<0>(value), std::get<1>(value));
+    }
+
+    static std::tuple<Pose3f, std::optional<Twist3f>> parseEvent(const ASensorEvent& event) {
+        // TODO(ytai): Add more types.
+        switch (event.type) {
+            case ASENSOR_TYPE_ROTATION_VECTOR:
+            case ASENSOR_TYPE_GAME_ROTATION_VECTOR: {
+                Eigen::Quaternionf quat(event.data[3], event.data[0], event.data[1], event.data[2]);
+                // Adapt to different frame convention.
+                quat *= rotateX(-M_PI_2);
+                return std::make_tuple(Pose3f(quat), std::optional<Twist3f>());
+            }
+
+            default:
+                ALOGE("Unsupported sensor type: %" PRId32, event.type);
+                return std::make_tuple(Pose3f(), std::optional<Twist3f>());
+        }
+    }
+};
+
+}  // namespace
+
+std::unique_ptr<SensorPoseProvider> SensorPoseProvider::create(const char* packageName,
+                                                               Listener* listener) {
+    return SensorPoseProviderImpl::create(packageName, listener);
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/TestUtil.h b/media/libheadtracking/TestUtil.h
new file mode 100644
index 0000000..4636d86
--- /dev/null
+++ b/media/libheadtracking/TestUtil.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <gtest/gtest.h>
+
+#include "media/Pose.h"
+#include "media/Twist.h"
+
+namespace {
+
+constexpr float kPoseComparisonPrecision = 1e-5;
+
+}  // namespace
+
+// These specializations make {EXPECT,ASSERT}_{EQ,NE} work correctly for Pose3f, Twist3f, Vector3f
+// and Quaternionf.
+namespace testing {
+namespace internal {
+
+template <>
+inline AssertionResult CmpHelperEQ<android::media::Pose3f, android::media::Pose3f>(
+        const char* lhs_expression, const char* rhs_expression, const android::media::Pose3f& lhs,
+        const android::media::Pose3f& rhs) {
+    if (lhs.isApprox(rhs, kPoseComparisonPrecision)) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperNE<android::media::Pose3f, android::media::Pose3f>(
+        const char* lhs_expression, const char* rhs_expression, const android::media::Pose3f& lhs,
+        const android::media::Pose3f& rhs) {
+    if (!lhs.isApprox(rhs, kPoseComparisonPrecision)) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperEQ<android::media::Twist3f, android::media::Twist3f>(
+        const char* lhs_expression, const char* rhs_expression, const android::media::Twist3f& lhs,
+        const android::media::Twist3f& rhs) {
+    if (lhs.isApprox(rhs, kPoseComparisonPrecision)) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperNE<android::media::Twist3f, android::media::Twist3f>(
+        const char* lhs_expression, const char* rhs_expression, const android::media::Twist3f& lhs,
+        const android::media::Twist3f& rhs) {
+    if (!lhs.isApprox(rhs, kPoseComparisonPrecision)) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperEQ<Eigen::Vector3f, Eigen::Vector3f>(const char* lhs_expression,
+                                                                     const char* rhs_expression,
+                                                                     const Eigen::Vector3f& lhs,
+                                                                     const Eigen::Vector3f& rhs) {
+    if (lhs.isApprox(rhs)) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperNE<Eigen::Vector3f, Eigen::Vector3f>(const char* lhs_expression,
+                                                                     const char* rhs_expression,
+                                                                     const Eigen::Vector3f& lhs,
+                                                                     const Eigen::Vector3f& rhs) {
+    if (!lhs.isApprox(rhs)) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperEQ<Eigen::Quaternionf, Eigen::Quaternionf>(
+        const char* lhs_expression, const char* rhs_expression, const Eigen::Quaternionf& lhs,
+        const Eigen::Quaternionf& rhs) {
+    // Negating the coefs results in an equivalent quaternion.
+    if (lhs.isApprox(rhs) || lhs.isApprox(Eigen::Quaternionf(-rhs.coeffs()))) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+template <>
+inline AssertionResult CmpHelperNE<Eigen::Quaternionf, Eigen::Quaternionf>(
+        const char* lhs_expression, const char* rhs_expression, const Eigen::Quaternionf& lhs,
+        const Eigen::Quaternionf& rhs) {
+    // Negating the coefs results in an equivalent quaternion.
+    if (!(lhs.isApprox(rhs) || lhs.isApprox(Eigen::Quaternionf(-rhs.coeffs())))) {
+        return AssertionSuccess();
+    }
+
+    return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+}
+
+}  // namespace internal
+}  // namespace testing
diff --git a/media/libheadtracking/Twist-test.cpp b/media/libheadtracking/Twist-test.cpp
new file mode 100644
index 0000000..7984e1e
--- /dev/null
+++ b/media/libheadtracking/Twist-test.cpp
@@ -0,0 +1,67 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "media/Twist.h"
+
+#include <gtest/gtest.h>
+
+#include "QuaternionUtil.h"
+#include "TestUtil.h"
+
+using Eigen::Quaternionf;
+using Eigen::Vector3f;
+
+namespace android {
+namespace media {
+namespace {
+
+TEST(Twist, DefaultCtor) {
+    Twist3f twist;
+    EXPECT_EQ(twist.translationalVelocity(), Vector3f::Zero());
+    EXPECT_EQ(twist.rotationalVelocity(), Vector3f::Zero());
+    EXPECT_FLOAT_EQ(twist.scalarRotationalVelocity(), 0);
+    EXPECT_FLOAT_EQ(twist.scalarTranslationalVelocity(), 0);
+}
+
+TEST(Twist, FullCtor) {
+    Vector3f rot{1, 2, 3};
+    Vector3f trans{4, 5, 6};
+    Twist3f twist(trans, rot);
+    EXPECT_EQ(twist.translationalVelocity(), trans);
+    EXPECT_EQ(twist.rotationalVelocity(), rot);
+    EXPECT_FLOAT_EQ(twist.scalarRotationalVelocity(), std::sqrt(14.f));
+    EXPECT_FLOAT_EQ(twist.scalarTranslationalVelocity(), std::sqrt(77.f));
+}
+
+TEST(Twist, Integrate) {
+    Vector3f trans{1, 2, 3};
+    // 45 deg/sec around Z.
+    Vector3f rot{0, 0, M_PI_4};
+    Twist3f twist(trans, rot);
+    Pose3f pose = integrate(twist, 2.f);
+
+    EXPECT_EQ(pose, Pose3f(Vector3f{2, 4, 6}, rotateZ(M_PI_2)));
+}
+
+TEST(Twist, Differentiate) {
+    Pose3f pose(Vector3f{2, 4, 6}, rotateZ(M_PI_2));
+    Twist3f twist = differentiate(pose, 2.f);
+    EXPECT_EQ(twist, Twist3f(Vector3f(1, 2, 3), Vector3f(0, 0, M_PI_4)));
+}
+
+}  // namespace
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/Twist.cpp b/media/libheadtracking/Twist.cpp
new file mode 100644
index 0000000..664c4d5
--- /dev/null
+++ b/media/libheadtracking/Twist.cpp
@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "media/Twist.h"
+
+#include "QuaternionUtil.h"
+
+namespace android {
+namespace media {
+
+Pose3f integrate(const Twist3f& twist, float dt) {
+    Eigen::Vector3f translation = twist.translationalVelocity() * dt;
+    Eigen::Vector3f rotationVector = twist.rotationalVelocity() * dt;
+    return Pose3f(translation, rotationVectorToQuaternion(rotationVector));
+}
+
+Twist3f differentiate(const Pose3f& pose, float dt) {
+    Eigen::Vector3f translationalVelocity = pose.translation() / dt;
+    Eigen::Vector3f rotationalVelocity = quaternionToRotationVector(pose.rotation()) / dt;
+    return Twist3f(translationalVelocity, rotationalVelocity);
+}
+
+std::ostream& operator<<(std::ostream& os, const Twist3f& twist) {
+    os << "translation: " << twist.translationalVelocity().transpose()
+       << " rotation vector: " << twist.rotationalVelocity().transpose();
+    return os;
+}
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/include/media/HeadTrackingMode.h b/media/libheadtracking/include/media/HeadTrackingMode.h
new file mode 100644
index 0000000..38496e8
--- /dev/null
+++ b/media/libheadtracking/include/media/HeadTrackingMode.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+namespace android {
+namespace media {
+
+/**
+ * Mode of head-tracking.
+ */
+enum class HeadTrackingMode {
+    /** No head-tracking - screen-to-head pose is assumed to be identity. */
+    STATIC,
+    /** Head tracking enabled - world-to-screen pose is assumed to be identity. */
+    WORLD_RELATIVE,
+    /** Full screen-to-head tracking enabled. */
+    SCREEN_RELATIVE,
+};
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/include/media/HeadTrackingProcessor.h b/media/libheadtracking/include/media/HeadTrackingProcessor.h
new file mode 100644
index 0000000..9fea273
--- /dev/null
+++ b/media/libheadtracking/include/media/HeadTrackingProcessor.h
@@ -0,0 +1,105 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <limits>
+
+#include "HeadTrackingMode.h"
+#include "Pose.h"
+#include "Twist.h"
+
+namespace android {
+namespace media {
+
+/**
+ * Main entry-point for this library.
+ * This interfaces encompasses all the processing required for determining the head-to-stage pose
+ * used for audio virtualization.
+ * The usage involves periodic setting of the inputs, calling calculate() and obtaining the outputs.
+ * This class is not thread-safe, but thread-compatible.
+ */
+class HeadTrackingProcessor {
+  public:
+    virtual ~HeadTrackingProcessor() = default;
+
+    struct Options {
+        float maxTranslationalVelocity = std::numeric_limits<float>::infinity();
+        float maxRotationalVelocity = std::numeric_limits<float>::infinity();
+        float translationalDriftTimeConstant = std::numeric_limits<float>::infinity();
+        float rotationalDriftTimeConstant = std::numeric_limits<float>::infinity();
+        int64_t freshnessTimeout = std::numeric_limits<int64_t>::max();
+        float predictionDuration = 0;
+    };
+
+    /** Sets the desired head-tracking mode. */
+    virtual void setDesiredMode(HeadTrackingMode mode) = 0;
+
+    /**
+     * Sets the world-to-head pose and head twist (velocity).
+     * headTwist is given in the head coordinate frame.
+     */
+    virtual void setWorldToHeadPose(int64_t timestamp, const Pose3f& worldToHead,
+                                    const Twist3f& headTwist) = 0;
+
+    /**
+     * Sets the world-to-screen pose.
+     */
+    virtual void setWorldToScreenPose(int64_t timestamp, const Pose3f& worldToScreen) = 0;
+
+    /**
+     * Set the screen-to-stage pose, used in all modes.
+     */
+    virtual void setScreenToStagePose(const Pose3f& screenToStage) = 0;
+
+    /**
+     * Sets the display orientation.
+     * Orientation is expressed in the angle of rotation from the physical "up" side of the screen
+     * to the logical "up" side of the content displayed the screen. Counterclockwise angles, as
+     * viewed while facing the screen are positive.
+     */
+    virtual void setDisplayOrientation(float physicalToLogicalAngle) = 0;
+
+    /**
+     * Process all the previous inputs and update the outputs.
+     */
+    virtual void calculate(int64_t timestamp) = 0;
+
+    /**
+     * Get the aggregate head-to-stage pose (primary output of this module).
+     */
+    virtual Pose3f getHeadToStagePose() const = 0;
+
+    /**
+     * Get the actual head-tracking mode (which may deviate from the desired one as mentioned in the
+     * class documentation above).
+     */
+    virtual HeadTrackingMode getActualMode() const = 0;
+
+    /**
+     * This causes the current poses for both the head and/or screen to be considered "center".
+     */
+    virtual void recenter(bool recenterHead = true, bool recenterScreen = true) = 0;
+};
+
+/**
+ * Creates an instance featuring a default implementation of the HeadTrackingProcessor interface.
+ */
+std::unique_ptr<HeadTrackingProcessor> createHeadTrackingProcessor(
+        const HeadTrackingProcessor::Options& options,
+        HeadTrackingMode initialMode = HeadTrackingMode::STATIC);
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/include/media/Pose.h b/media/libheadtracking/include/media/Pose.h
new file mode 100644
index 0000000..e660bb9
--- /dev/null
+++ b/media/libheadtracking/include/media/Pose.h
@@ -0,0 +1,132 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <optional>
+#include <vector>
+#include <Eigen/Geometry>
+
+namespace android {
+namespace media {
+
+/**
+ * A 6-DoF pose.
+ * This class represents a proper rigid transformation (translation + rotation) between a reference
+ * frame and a target frame,
+ *
+ * See https://en.wikipedia.org/wiki/Six_degrees_of_freedom
+ */
+class Pose3f {
+  public:
+    /** Typical precision for isApprox comparisons. */
+    static constexpr float kDummyPrecision = 1e-5f;
+
+    Pose3f(const Eigen::Vector3f& translation, const Eigen::Quaternionf& rotation)
+        : mTranslation(translation), mRotation(rotation) {}
+
+    explicit Pose3f(const Eigen::Vector3f& translation)
+        : Pose3f(translation, Eigen::Quaternionf::Identity()) {}
+
+    explicit Pose3f(const Eigen::Quaternionf& rotation)
+        : Pose3f(Eigen::Vector3f::Zero(), rotation) {}
+
+    Pose3f() : Pose3f(Eigen::Vector3f::Zero(), Eigen::Quaternionf::Identity()) {}
+
+    Pose3f(const Pose3f& other) { *this = other; }
+
+    /**
+     * Create instance from a vector-of-floats representation.
+     * The vector is expected to have exactly 6 elements, where the first three are a translation
+     * vector and the last three are a rotation vector.
+     *
+     * Returns nullopt if the input vector is illegal.
+     */
+    static std::optional<Pose3f> fromVector(const std::vector<float>& vec);
+
+    /**
+     * Convert instance to a vector-of-floats representation.
+     * The vector will have exactly 6 elements, where the first three are a translation vector and
+     * the last three are a rotation vector.
+     */
+    std::vector<float> toVector() const;
+
+    Pose3f& operator=(const Pose3f& other) {
+        mTranslation = other.mTranslation;
+        mRotation = other.mRotation;
+        return *this;
+    }
+
+    Eigen::Vector3f translation() const { return mTranslation; };
+    Eigen::Quaternionf rotation() const { return mRotation; };
+
+    /**
+     * Reverses the reference and target frames.
+     */
+    Pose3f inverse() const {
+        Eigen::Quaternionf invRotation = mRotation.inverse();
+        return Pose3f(-(invRotation * translation()), invRotation);
+    }
+
+    /**
+     * Composes (chains) together two poses. By convention, this only makes sense if the target
+     * frame of the left-hand pose is the same the reference frame of the right-hand pose.
+     * Note that this operator is not commutative.
+     */
+    Pose3f operator*(const Pose3f& other) const {
+        Pose3f result = *this;
+        result *= other;
+        return result;
+    }
+
+    Pose3f& operator*=(const Pose3f& other) {
+        mTranslation += mRotation * other.mTranslation;
+        mRotation *= other.mRotation;
+        return *this;
+    }
+
+    /**
+     * This is an imprecise "fuzzy" comparison, which is only to be used for validity-testing
+     * purposes.
+     */
+    bool isApprox(const Pose3f& other, float prec = kDummyPrecision) const {
+        return (mTranslation - other.mTranslation).norm() < prec &&
+               // Quaternions are equivalent under sign inversion.
+               ((mRotation.coeffs() - other.mRotation.coeffs()).norm() < prec ||
+                (mRotation.coeffs() + other.mRotation.coeffs()).norm() < prec);
+    }
+
+  private:
+    Eigen::Vector3f mTranslation;
+    Eigen::Quaternionf mRotation;
+};
+
+/**
+ * Pretty-printer for Pose3f.
+ */
+std::ostream& operator<<(std::ostream& os, const Pose3f& pose);
+
+/**
+ * Move between the 'from' pose and the 'to' pose, while making sure velocity limits are enforced.
+ * If velocity limits are not violated, returns the 'to' pose and false.
+ * If velocity limits are violated, returns pose farthest along the path that can be reached within
+ * the limits, and true.
+ */
+std::tuple<Pose3f, bool> moveWithRateLimit(const Pose3f& from, const Pose3f& to, float t,
+                                           float maxTranslationalVelocity,
+                                           float maxRotationalVelocity);
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/include/media/SensorPoseProvider.h b/media/libheadtracking/include/media/SensorPoseProvider.h
new file mode 100644
index 0000000..3e9c107
--- /dev/null
+++ b/media/libheadtracking/include/media/SensorPoseProvider.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <chrono>
+#include <memory>
+#include <optional>
+
+#include <android/sensor.h>
+
+#include "Pose.h"
+#include "Twist.h"
+
+namespace android {
+namespace media {
+
+/**
+ * A utility providing streaming of pose data from motion sensors provided by the Sensor Framework.
+ *
+ * A live instance of this interface keeps around some resources required for accessing sensor
+ * readings (e.g. a thread and a queue). Those would be released when the instance is deleted.
+ *
+ * Once alive, individual sensors can be subscribed to using startSensor() and updates can be
+ * stopped via stopSensor(). Those two methods should not be called concurrently and correct usage
+ * is assumed.
+ */
+class SensorPoseProvider {
+  public:
+    static constexpr int32_t INVALID_HANDLE = ASENSOR_INVALID;
+
+    /**
+     * Interface for consuming pose-related sensor events.
+     *
+     * The listener will be provided with a stream of events, each including:
+     * - A handle of the sensor responsible for the event.
+     * - Timestamp.
+     * - Pose.
+     * - Optional twist (time-derivative of pose).
+     *
+     * Sensors having only orientation data will have the translation part of the pose set to
+     * identity.
+     *
+     * Events are delivered in a serialized manner (i.e. callbacks do not need to be reentrant).
+     * Callbacks should not block.
+     */
+    class Listener {
+      public:
+        virtual ~Listener() = default;
+
+        virtual void onPose(int64_t timestamp, int32_t handle, const Pose3f& pose,
+                            const std::optional<Twist3f>& twist) = 0;
+    };
+
+    /**
+     * Creates a new SensorPoseProvider instance.
+     * Events will be delivered to the listener as long as the returned instance is kept alive.
+     * @param packageName Client's package name.
+     * @param listener The listener that will get the events.
+     * @return The new instance, or nullptr in case of failure.
+     */
+    static std::unique_ptr<SensorPoseProvider> create(const char* packageName, Listener* listener);
+
+    virtual ~SensorPoseProvider() = default;
+
+    /**
+     * Start receiving pose updates from a given sensor.
+     * Attempting to start a sensor that has already been started results in undefined behavior.
+     * @param sensor The sensor to subscribe to.
+     * @param samplingPeriod Sampling interval, in microseconds. Actual rate might be slightly
+     * different.
+     * @return The sensor handle, which can be later used for stopSensor(). INVALID_HANDLE would be
+     * returned in case of error. This is guaranteed to be the same handle as the one returned by
+     * ASensor_getHandle(sensor).
+     */
+    virtual int32_t startSensor(const ASensor* sensor,
+                                std::chrono::microseconds samplingPeriod) = 0;
+
+    /**
+     * Stop a sensor, previously started with startSensor(). It is not required to stop all sensors
+     * before deleting the SensorPoseProvider instance.
+     * @param handle The sensor handle, as returned from startSensor().
+     */
+    virtual void stopSensor(int32_t handle) = 0;
+};
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libheadtracking/include/media/Twist.h b/media/libheadtracking/include/media/Twist.h
new file mode 100644
index 0000000..e2fc203
--- /dev/null
+++ b/media/libheadtracking/include/media/Twist.h
@@ -0,0 +1,84 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <Eigen/Geometry>
+
+#include "Pose.h"
+
+namespace android {
+namespace media {
+
+/**
+ * A 6-DoF twist.
+ * This class represents the translational and rotational velocity of a rigid object, typically
+ * relative to its own coordinate-frame.
+ * It is created by two 3-vectors, one representing linear motion per time-unit and the other, a
+ * rotation-vector in radians per time-unit (right-handed).
+ */
+class Twist3f {
+  public:
+    Twist3f(const Eigen::Vector3f& translationalVelocity, const Eigen::Vector3f& rotationalVelocity)
+        : mTranslationalVelocity(translationalVelocity), mRotationalVelocity(rotationalVelocity) {}
+
+    Twist3f() : Twist3f(Eigen::Vector3f::Zero(), Eigen::Vector3f::Zero()) {}
+
+    Twist3f(const Twist3f& other) { *this = other; }
+
+    Twist3f& operator=(const Twist3f& other) {
+        mTranslationalVelocity = other.mTranslationalVelocity;
+        mRotationalVelocity = other.mRotationalVelocity;
+        return *this;
+    }
+
+    Eigen::Vector3f translationalVelocity() const { return mTranslationalVelocity; }
+    Eigen::Vector3f rotationalVelocity() const { return mRotationalVelocity; }
+
+    float scalarTranslationalVelocity() const { return mTranslationalVelocity.norm(); }
+    float scalarRotationalVelocity() const { return mRotationalVelocity.norm(); }
+
+    bool isApprox(const Twist3f& other,
+                  float prec = Eigen::NumTraits<float>::dummy_precision()) const {
+        return mTranslationalVelocity.isApprox(other.mTranslationalVelocity, prec) &&
+               mRotationalVelocity.isApprox(other.mRotationalVelocity, prec);
+    }
+
+  private:
+    Eigen::Vector3f mTranslationalVelocity;
+    Eigen::Vector3f mRotationalVelocity;
+};
+
+/**
+ * Integrate a twist over time to obtain a pose.
+ * dt is the time over which to integration.
+ * The resulting pose represents the transformation between the starting point and the ending point
+ * of the motion over the time period.
+ */
+Pose3f integrate(const Twist3f& twist, float dt);
+
+/**
+ * Differentiate pose to obtain a twist.
+ * dt is the time of the motion between the reference and the target frames of the pose.
+ */
+Twist3f differentiate(const Pose3f& pose, float dt);
+
+/**
+ * Pretty-printer for twist.
+ */
+std::ostream& operator<<(std::ostream& os, const Twist3f& twist);
+
+}  // namespace media
+}  // namespace android
diff --git a/media/libmediahelper/TypeConverter.cpp b/media/libmediahelper/TypeConverter.cpp
index d3a517f..97b5b95 100644
--- a/media/libmediahelper/TypeConverter.cpp
+++ b/media/libmediahelper/TypeConverter.cpp
@@ -50,6 +50,8 @@
     MAKE_STRING_FROM_ENUM(AUDIO_FLAG_MUTE_HAPTIC),
     MAKE_STRING_FROM_ENUM(AUDIO_FLAG_NO_SYSTEM_CAPTURE),
     MAKE_STRING_FROM_ENUM(AUDIO_FLAG_CAPTURE_PRIVATE),
+    MAKE_STRING_FROM_ENUM(AUDIO_FLAG_CONTENT_SPATIALIZED),
+    MAKE_STRING_FROM_ENUM(AUDIO_FLAG_NEVER_SPATIALIZE),
     TERMINATOR
 };
 
diff --git a/media/libstagefright/ACodec.cpp b/media/libstagefright/ACodec.cpp
index 1aa1848..1e64538 100644
--- a/media/libstagefright/ACodec.cpp
+++ b/media/libstagefright/ACodec.cpp
@@ -5395,21 +5395,21 @@
                             err = mOMXNode->getParameter(
                                     (OMX_INDEXTYPE)OMX_IndexParamAudioAndroidAacDrcPresentation,
                                     &presentation, sizeof(presentation));
-                            if (err != OK) {
-                                return err;
+                            if (err == OK) {
+                                notify->setInt32("aac-encoded-target-level",
+                                                 presentation.nEncodedTargetLevel);
+                                notify->setInt32("aac-drc-cut-level", presentation.nDrcCut);
+                                notify->setInt32("aac-drc-boost-level", presentation.nDrcBoost);
+                                notify->setInt32("aac-drc-heavy-compression",
+                                                 presentation.nHeavyCompression);
+                                notify->setInt32("aac-target-ref-level",
+                                                 presentation.nTargetReferenceLevel);
+                                notify->setInt32("aac-drc-effect-type",
+                                                 presentation.nDrcEffectType);
+                                notify->setInt32("aac-drc-album-mode", presentation.nDrcAlbumMode);
+                                notify->setInt32("aac-drc-output-loudness",
+                                                 presentation.nDrcOutputLoudness);
                             }
-                            notify->setInt32("aac-encoded-target-level",
-                                             presentation.nEncodedTargetLevel);
-                            notify->setInt32("aac-drc-cut-level", presentation.nDrcCut);
-                            notify->setInt32("aac-drc-boost-level", presentation.nDrcBoost);
-                            notify->setInt32("aac-drc-heavy-compression",
-                                             presentation.nHeavyCompression);
-                            notify->setInt32("aac-target-ref-level",
-                                             presentation.nTargetReferenceLevel);
-                            notify->setInt32("aac-drc-effect-type", presentation.nDrcEffectType);
-                            notify->setInt32("aac-drc-album-mode", presentation.nDrcAlbumMode);
-                            notify->setInt32("aac-drc-output-loudness",
-                                             presentation.nDrcOutputLoudness);
                         }
                     }
                     break;
diff --git a/media/libstagefright/FrameDecoder.cpp b/media/libstagefright/FrameDecoder.cpp
index b513e45..a241e29 100644
--- a/media/libstagefright/FrameDecoder.cpp
+++ b/media/libstagefright/FrameDecoder.cpp
@@ -796,8 +796,16 @@
     if (overrideMeta == NULL) {
         // check if we're dealing with a tiled heif
         int32_t tileWidth, tileHeight, gridRows, gridCols;
+        int32_t widthColsProduct = 0;
+        int32_t heightRowsProduct = 0;
         if (findGridInfo(trackMeta(), &tileWidth, &tileHeight, &gridRows, &gridCols)) {
-            if (mWidth <= tileWidth * gridCols && mHeight <= tileHeight * gridRows) {
+            if (__builtin_mul_overflow(tileWidth, gridCols, &widthColsProduct) ||
+                    __builtin_mul_overflow(tileHeight, gridRows, &heightRowsProduct)) {
+                ALOGE("Multiplication overflowed Grid size: %dx%d, Picture size: %dx%d",
+                        gridCols, gridRows, tileWidth, tileHeight);
+                return nullptr;
+            }
+            if (mWidth <= widthColsProduct && mHeight <= heightRowsProduct) {
                 ALOGV("grid: %dx%d, tile size: %dx%d, picture size: %dx%d",
                         gridCols, gridRows, tileWidth, tileHeight, mWidth, mHeight);
 
diff --git a/media/ndk/NdkMediaFormat.cpp b/media/ndk/NdkMediaFormat.cpp
index 51f6c78..69ab242 100644
--- a/media/ndk/NdkMediaFormat.cpp
+++ b/media/ndk/NdkMediaFormat.cpp
@@ -354,6 +354,11 @@
 EXPORT const char* AMEDIAFORMAT_KEY_MIME = "mime";
 EXPORT const char* AMEDIAFORMAT_KEY_MPEG_USER_DATA = "mpeg-user-data";
 EXPORT const char* AMEDIAFORMAT_KEY_MPEG2_STREAM_HEADER = "mpeg2-stream-header";
+EXPORT const char* AMEDIAFORMAT_KEY_MPEGH_COMPATIBLE_SETS = "mpegh-compatible-sets";
+EXPORT const char* AMEDIAFORMAT_KEY_MPEGH_PROFILE_LEVEL_INDICATION =
+        "mpegh-profile-level-indication";
+EXPORT const char* AMEDIAFORMAT_KEY_MPEGH_REFERENCE_CHANNEL_LAYOUT =
+        "mpegh-reference-channel-layout";
 EXPORT const char* AMEDIAFORMAT_KEY_OPERATING_RATE = "operating-rate";
 EXPORT const char* AMEDIAFORMAT_KEY_PCM_ENCODING = "pcm-encoding";
 EXPORT const char* AMEDIAFORMAT_KEY_PRIORITY = "priority";
diff --git a/media/ndk/include/media/NdkMediaFormat.h b/media/ndk/include/media/NdkMediaFormat.h
index fbd855d..2d2fcc0 100644
--- a/media/ndk/include/media/NdkMediaFormat.h
+++ b/media/ndk/include/media/NdkMediaFormat.h
@@ -320,6 +320,34 @@
 extern const char* AMEDIAFORMAT_VIDEO_QP_P_MAX __INTRODUCED_IN(31);
 extern const char* AMEDIAFORMAT_VIDEO_QP_P_MIN __INTRODUCED_IN(31);
 
+/**
+ * MPEG-H audio profile and level compatibility.
+ *
+ * See FDAmd_2 of ISO_IEC_23008-3;2019 MHAProfileAndLevelCompatibilitySetBox.
+ *
+ * Available since API level 32.
+ */
+extern const char* AMEDIAFORMAT_KEY_MPEGH_COMPATIBLE_SETS __INTRODUCED_IN(32);
+
+/**
+ * MPEG-H audio profile level indication.
+ *
+ * See ISO_IEC_23008-3;2019 MHADecoderConfigurationRecord mpegh3daProfileLevelIndication.
+ *
+ * Available since API level 32.
+ */
+extern const char* AMEDIAFORMAT_KEY_MPEGH_PROFILE_LEVEL_INDICATION __INTRODUCED_IN(32);
+
+/**
+ * MPEG-H audio reference channel layout.
+ *
+ * See ISO_IEC_23008-3;2019 MHADecoderConfigurationRecord referenceChannelLayout
+ * and ISO_IEC_23001‐8 ChannelConfiguration value.
+ *
+ * Available since API level 32.
+ */
+extern const char* AMEDIAFORMAT_KEY_MPEGH_REFERENCE_CHANNEL_LAYOUT __INTRODUCED_IN(32);
+
 __END_DECLS
 
 #endif // _NDK_MEDIA_FORMAT_H
diff --git a/media/ndk/libmediandk.map.txt b/media/ndk/libmediandk.map.txt
index 7e9e57e..6f275c7 100644
--- a/media/ndk/libmediandk.map.txt
+++ b/media/ndk/libmediandk.map.txt
@@ -126,6 +126,9 @@
     AMEDIAFORMAT_KEY_MIME; # var introduced=21
     AMEDIAFORMAT_KEY_MPEG_USER_DATA; # var introduced=28
     AMEDIAFORMAT_KEY_MPEG2_STREAM_HEADER; # var introduced=29
+    AMEDIAFORMAT_KEY_MPEGH_COMPATIBLE_SETS; # var introduced=32
+    AMEDIAFORMAT_KEY_MPEGH_PROFILE_LEVEL_INDICATION; # var introduced=32
+    AMEDIAFORMAT_KEY_MPEGH_REFERENCE_CHANNEL_LAYOUT; # var introduced=32
     AMEDIAFORMAT_KEY_OPERATING_RATE; # var introduced=28
     AMEDIAFORMAT_KEY_PCM_BIG_ENDIAN; # var introduced=29
     AMEDIAFORMAT_KEY_PCM_ENCODING; # var introduced=28
diff --git a/media/utils/Android.bp b/media/utils/Android.bp
index bfe73d5..73c4e3b 100644
--- a/media/utils/Android.bp
+++ b/media/utils/Android.bp
@@ -81,6 +81,36 @@
     export_include_dirs: ["include"],
 }
 
+cc_library {
+    name: "libmediautils_vendor",
+    vendor_available: true,  // required for platform/hardware/interfaces
+    srcs: [
+        "MemoryLeakTrackUtil.cpp",
+    ],
+
+    cflags: [
+        "-Wall",
+        "-Wextra",
+        "-Werror",
+    ],
+    shared_libs: [
+        "liblog",
+        "libutils",
+    ],
+
+    static_libs: [
+        "libc_malloc_debug_backtrace",
+    ],
+
+    header_libs: [
+        "bionic_libc_platform_headers",
+    ],
+
+    local_include_dirs: ["include"],
+    export_include_dirs: ["include"],
+}
+
+
 cc_library_headers {
     name: "libmediautils_headers",
     vendor_available: true,  // required for platform/hardware/interfaces
diff --git a/services/audioflinger/AudioFlinger.cpp b/services/audioflinger/AudioFlinger.cpp
index 65a163f..676bf78 100644
--- a/services/audioflinger/AudioFlinger.cpp
+++ b/services/audioflinger/AudioFlinger.cpp
@@ -64,6 +64,7 @@
 #include <system/audio_effects/effect_ns.h>
 #include <system/audio_effects/effect_aec.h>
 #include <system/audio_effects/effect_hapticgenerator.h>
+#include <system/audio_effects/effect_spatializer.h>
 
 #include <audio_utils/primitives.h>
 
@@ -336,11 +337,11 @@
 }
 
 // getDefaultVibratorInfo_l must be called with AudioFlinger lock held.
-const media::AudioVibratorInfo* AudioFlinger::getDefaultVibratorInfo_l() {
+std::optional<media::AudioVibratorInfo> AudioFlinger::getDefaultVibratorInfo_l() {
     if (mAudioVibratorInfos.empty()) {
-        return nullptr;
+        return {};
     }
-    return &mAudioVibratorInfos.front();
+    return mAudioVibratorInfos.front();
 }
 
 AudioFlinger::~AudioFlinger()
@@ -695,7 +696,7 @@
         // dump all hardware devs
         for (size_t i = 0; i < mAudioHwDevs.size(); i++) {
             sp<DeviceHalInterface> dev = mAudioHwDevs.valueAt(i)->hwDevice();
-            dev->dump(fd);
+            dev->dump(fd, args);
         }
 
         mPatchPanel.dump(fd);
@@ -2455,6 +2456,10 @@
         ThreadBase *thread = (ThreadBase *)mRecordThreads.valueAt(i).get();
         thread->systemReady();
     }
+    for (size_t i = 0; i < mMmapThreads.size(); i++) {
+        ThreadBase *thread = (ThreadBase *)mMmapThreads.valueAt(i).get();
+        thread->systemReady();
+    }
     return NO_ERROR;
 }
 
@@ -2501,7 +2506,8 @@
 
 sp<AudioFlinger::ThreadBase> AudioFlinger::openOutput_l(audio_module_handle_t module,
                                                         audio_io_handle_t *output,
-                                                        audio_config_t *config,
+                                                        audio_config_t *halConfig,
+                                                        audio_config_base_t *mixerConfig __unused,
                                                         audio_devices_t deviceType,
                                                         const String8& address,
                                                         audio_output_flags_t flags)
@@ -2529,16 +2535,16 @@
         // Check only for Normal Mixing mode
         if (kEnableExtendedPrecision) {
             // Specify format (uncomment one below to choose)
-            //config->format = AUDIO_FORMAT_PCM_FLOAT;
-            //config->format = AUDIO_FORMAT_PCM_24_BIT_PACKED;
-            //config->format = AUDIO_FORMAT_PCM_32_BIT;
-            //config->format = AUDIO_FORMAT_PCM_8_24_BIT;
-            // ALOGV("openOutput_l() upgrading format to %#08x", config->format);
+            //halConfig->format = AUDIO_FORMAT_PCM_FLOAT;
+            //halConfig->format = AUDIO_FORMAT_PCM_24_BIT_PACKED;
+            //halConfig->format = AUDIO_FORMAT_PCM_32_BIT;
+            //halConfig->format = AUDIO_FORMAT_PCM_8_24_BIT;
+            // ALOGV("openOutput_l() upgrading format to %#08x", halConfig->format);
         }
         if (kEnableExtendedChannels) {
             // Specify channel mask (uncomment one below to choose)
-            //config->channel_mask = audio_channel_out_mask_from_count(4);  // for USB 4ch
-            //config->channel_mask = audio_channel_mask_from_representation_and_bits(
+            //halConfig->channel_mask = audio_channel_out_mask_from_count(4);  // for USB 4ch
+            //halConfig->channel_mask = audio_channel_mask_from_representation_and_bits(
             //        AUDIO_CHANNEL_REPRESENTATION_INDEX, (1 << 4) - 1);  // another 4ch example
         }
     }
@@ -2549,7 +2555,7 @@
             *output,
             deviceType,
             flags,
-            config,
+            halConfig,
             address.string());
 
     mHardwareStatus = AUDIO_HW_IDLE;
@@ -2564,13 +2570,20 @@
             return thread;
         } else {
             sp<PlaybackThread> thread;
-            if (flags & AUDIO_OUTPUT_FLAG_COMPRESS_OFFLOAD) {
+            //TODO: b/193496180 use spatializer flag at audio HAL when available
+            if (flags == (audio_output_flags_t)(AUDIO_OUTPUT_FLAG_FAST
+                                                    | AUDIO_OUTPUT_FLAG_DEEP_BUFFER)) {
+                thread = new SpatializerThread(this, outputStream, *output,
+                                                    mSystemReady, mixerConfig);
+                ALOGD("openOutput_l() created virtualizer output: ID %d thread %p",
+                      *output, thread.get());
+            } else if (flags & AUDIO_OUTPUT_FLAG_COMPRESS_OFFLOAD) {
                 thread = new OffloadThread(this, outputStream, *output, mSystemReady);
                 ALOGV("openOutput_l() created offload output: ID %d thread %p",
                       *output, thread.get());
             } else if ((flags & AUDIO_OUTPUT_FLAG_DIRECT)
-                    || !isValidPcmSinkFormat(config->format)
-                    || !isValidPcmSinkChannelMask(config->channel_mask)) {
+                    || !isValidPcmSinkFormat(halConfig->format)
+                    || !isValidPcmSinkChannelMask(halConfig->channel_mask)) {
                 thread = new DirectOutputThread(this, outputStream, *output, mSystemReady);
                 ALOGV("openOutput_l() created direct output: ID %d thread %p",
                       *output, thread.get());
@@ -2597,8 +2610,10 @@
 {
     audio_module_handle_t module = VALUE_OR_RETURN_STATUS(
             aidl2legacy_int32_t_audio_module_handle_t(request.module));
-    audio_config_t config = VALUE_OR_RETURN_STATUS(
-            aidl2legacy_AudioConfig_audio_config_t(request.config));
+    audio_config_t halConfig = VALUE_OR_RETURN_STATUS(
+            aidl2legacy_AudioConfig_audio_config_t(request.halConfig));
+    audio_config_base_t mixerConfig = VALUE_OR_RETURN_STATUS(
+            aidl2legacy_AudioConfigBase_audio_config_base_t(request.mixerConfig));
     sp<DeviceDescriptorBase> device = VALUE_OR_RETURN_STATUS(
             aidl2legacy_DeviceDescriptorBase(request.device));
     audio_output_flags_t flags = VALUE_OR_RETURN_STATUS(
@@ -2611,9 +2626,9 @@
               "Channels %#x, flags %#x",
               this, module,
               device->toString().c_str(),
-              config.sample_rate,
-              config.format,
-              config.channel_mask,
+              halConfig.sample_rate,
+              halConfig.format,
+              halConfig.channel_mask,
               flags);
 
     audio_devices_t deviceType = device->type();
@@ -2625,7 +2640,8 @@
 
     Mutex::Autolock _l(mLock);
 
-    sp<ThreadBase> thread = openOutput_l(module, &output, &config, deviceType, address, flags);
+    sp<ThreadBase> thread = openOutput_l(module, &output, &halConfig,
+            &mixerConfig, deviceType, address, flags);
     if (thread != 0) {
         if ((flags & AUDIO_OUTPUT_FLAG_MMAP_NOIRQ) == 0) {
             PlaybackThread *playbackThread = (PlaybackThread *)thread.get();
@@ -2650,7 +2666,8 @@
             mmapThread->ioConfigChanged(AUDIO_OUTPUT_OPENED);
         }
         response->output = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_io_handle_t_int32_t(output));
-        response->config = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_t_AudioConfig(config));
+        response->config =
+                VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_t_AudioConfig(halConfig));
         response->latencyMs = VALUE_OR_RETURN_STATUS(convertIntegral<int32_t>(latencyMs));
         response->flags = VALUE_OR_RETURN_STATUS(
                 legacy2aidl_audio_output_flags_t_int32_t_mask(flags));
@@ -3729,6 +3746,15 @@
             goto Exit;
         }
 
+        // Only audio policy service can create a spatializer effect
+        if ((memcmp(&descOut.type, FX_IID_SPATIALIZER, sizeof(effect_uuid_t)) == 0) &&
+            (callingUid != AID_AUDIOSERVER || currentPid != getpid())) {
+            ALOGW("%s: attempt to create a spatializer effect from uid/pid %d/%d",
+                    __func__, callingUid, currentPid);
+            lStatus = PERMISSION_DENIED;
+            goto Exit;
+        }
+
         if (io == AUDIO_IO_HANDLE_NONE && sessionId == AUDIO_SESSION_OUTPUT_MIX) {
             // if the output returned by getOutputForEffect() is removed before we lock the
             // mutex below, the call to checkPlaybackThread_l(io) below will detect it
@@ -3744,7 +3770,7 @@
             ALOGV("%s device type %#x address %s", __func__, device.mType, device.getAddress());
             handle = mDeviceEffectManager.createEffect_l(
                     &descOut, device, client, effectClient, mPatchPanel.patches_l(),
-                    &enabledOut, &lStatus, probe);
+                    &enabledOut, &lStatus, probe, request.notifyFramesProcessed);
             if (lStatus != NO_ERROR && lStatus != ALREADY_EXISTS) {
                 // remove local strong reference to Client with mClientLock held
                 Mutex::Autolock _cl(mClientLock);
@@ -3797,7 +3823,8 @@
                 io = mPlaybackThreads.keyAt(0);
             }
             ALOGV("createEffect() got io %d for effect %s", io, descOut.name);
-        } else if (checkPlaybackThread_l(io) != nullptr) {
+        } else if (checkPlaybackThread_l(io) != nullptr
+                        && sessionId != AUDIO_SESSION_OUTPUT_STAGE) {
             // allow only one effect chain per sessionId on mPlaybackThreads.
             for (size_t i = 0; i < mPlaybackThreads.size(); i++) {
                 const audio_io_handle_t checkIo = mPlaybackThreads.keyAt(i);
@@ -3863,7 +3890,8 @@
             }
         }
         handle = thread->createEffect_l(client, effectClient, priority, sessionId,
-                                        &descOut, &enabledOut, &lStatus, pinned, probe);
+                                        &descOut, &enabledOut, &lStatus, pinned, probe,
+                                        request.notifyFramesProcessed);
         if (lStatus != NO_ERROR && lStatus != ALREADY_EXISTS) {
             // remove local strong reference to Client with mClientLock held
             Mutex::Autolock _cl(mClientLock);
@@ -4178,6 +4206,7 @@
         case TransactionCode::LIST_AUDIO_PATCHES:
         case TransactionCode::SET_AUDIO_PORT_CONFIG:
         case TransactionCode::SET_RECORD_SILENCED:
+        case TransactionCode::AUDIO_POLICY_READY:
             ALOGW("%s: transaction %d received from PID %d",
                   __func__, code, IPCThreadState::self()->getCallingPid());
             // return status only for non void methods
diff --git a/services/audioflinger/AudioFlinger.h b/services/audioflinger/AudioFlinger.h
index fff61f8..8fcd6e4 100644
--- a/services/audioflinger/AudioFlinger.h
+++ b/services/audioflinger/AudioFlinger.h
@@ -269,6 +269,9 @@
 
     /* Indicate JAVA services are ready (scheduling, power management ...) */
     virtual status_t systemReady();
+    virtual status_t audioPolicyReady() { mAudioPolicyReady.store(true); return NO_ERROR; }
+            bool isAudioPolicyReady() const { return mAudioPolicyReady.load(); }
+
 
     virtual status_t getMicrophones(std::vector<media::MicrophoneInfo> *microphones);
 
@@ -309,7 +312,7 @@
     void updateDownStreamPatches_l(const struct audio_patch *patch,
                                    const std::set<audio_io_handle_t> streams);
 
-    const media::AudioVibratorInfo* getDefaultVibratorInfo_l();
+    std::optional<media::AudioVibratorInfo> getDefaultVibratorInfo_l();
 
 private:
     // FIXME The 400 is temporarily too high until a leak of writers in media.log is fixed.
@@ -735,7 +738,8 @@
                                            const String8& outputDeviceAddress);
               sp<ThreadBase> openOutput_l(audio_module_handle_t module,
                                           audio_io_handle_t *output,
-                                          audio_config_t *config,
+                                          audio_config_t *halConfig,
+                                          audio_config_base_t *mixerConfig,
                                           audio_devices_t deviceType,
                                           const String8& address,
                                           audio_output_flags_t flags);
@@ -986,6 +990,7 @@
     DeviceEffectManager mDeviceEffectManager;
 
     bool       mSystemReady;
+    std::atomic_bool mAudioPolicyReady{};
 
     mediautils::UidInfo mUidInfo;
 
diff --git a/services/audioflinger/DeviceEffectManager.cpp b/services/audioflinger/DeviceEffectManager.cpp
index cecd52b..53ac5cb 100644
--- a/services/audioflinger/DeviceEffectManager.cpp
+++ b/services/audioflinger/DeviceEffectManager.cpp
@@ -77,7 +77,8 @@
         const std::map<audio_patch_handle_t, PatchPanel::Patch>& patches,
         int *enabled,
         status_t *status,
-        bool probe) {
+        bool probe,
+        bool notifyFramesProcessed) {
     sp<DeviceEffectProxy> effect;
     sp<EffectHandle> handle;
     status_t lStatus;
@@ -95,10 +96,12 @@
             effect = iter->second;
         } else {
             effect = new DeviceEffectProxy(device, mMyCallback,
-                    descriptor, mAudioFlinger.nextUniqueId(AUDIO_UNIQUE_ID_USE_EFFECT));
+                    descriptor, mAudioFlinger.nextUniqueId(AUDIO_UNIQUE_ID_USE_EFFECT),
+                    notifyFramesProcessed);
         }
         // create effect handle and connect it to effect module
-        handle = new EffectHandle(effect, client, effectClient, 0 /*priority*/);
+        handle = new EffectHandle(effect, client, effectClient, 0 /*priority*/,
+                                  notifyFramesProcessed);
         lStatus = handle->initCheck();
         if (lStatus == NO_ERROR) {
             lStatus = effect->addHandle(handle.get());
diff --git a/services/audioflinger/DeviceEffectManager.h b/services/audioflinger/DeviceEffectManager.h
index a05f5fe..5cfe2fb 100644
--- a/services/audioflinger/DeviceEffectManager.h
+++ b/services/audioflinger/DeviceEffectManager.h
@@ -37,7 +37,8 @@
                 const std::map<audio_patch_handle_t, PatchPanel::Patch>& patches,
                 int *enabled,
                 status_t *status,
-                bool probe);
+                bool probe,
+                bool notifyFramesProcessed);
     void createAudioPatch(audio_patch_handle_t handle, const PatchPanel::Patch& patch);
     void releaseAudioPatch(audio_patch_handle_t handle);
 
@@ -163,8 +164,13 @@
     bool isOffloadOrMmap() const override { return false; }
 
     uint32_t  sampleRate() const override { return 0; }
-    audio_channel_mask_t channelMask() const override { return AUDIO_CHANNEL_NONE; }
-    uint32_t channelCount() const override { return 0; }
+    audio_channel_mask_t inChannelMask(int id __unused) const override {
+        return AUDIO_CHANNEL_NONE;
+    }
+    uint32_t inChannelCount(int id __unused) const override { return 0; }
+    audio_channel_mask_t outChannelMask() const override { return AUDIO_CHANNEL_NONE; }
+    uint32_t outChannelCount() const override { return 0; }
+
     audio_channel_mask_t hapticChannelMask() const override { return AUDIO_CHANNEL_NONE; }
     size_t    frameCount() const override  { return 0; }
     uint32_t  latency() const override  { return 0; }
@@ -190,6 +196,10 @@
 
     wp<EffectChain> chain() const override { return nullptr; }
 
+    bool isAudioPolicyReady() const override {
+        return mManager.audioFlinger().isAudioPolicyReady();
+    }
+
     int newEffectId() { return mManager.audioFlinger().nextUniqueId(AUDIO_UNIQUE_ID_USE_EFFECT); }
 
     status_t addEffectToHal(audio_port_handle_t deviceId,
diff --git a/services/audioflinger/Effects.cpp b/services/audioflinger/Effects.cpp
index b267d88..b80943e 100644
--- a/services/audioflinger/Effects.cpp
+++ b/services/audioflinger/Effects.cpp
@@ -242,6 +242,12 @@
 
     {
         Mutex::Autolock _l(mLock);
+
+        if ((isInternal_l() && !mPolicyRegistered)
+                || !getCallback()->isAudioPolicyReady()) {
+            return NO_ERROR;
+        }
+
         // register effect when first handle is attached and unregister when last handle is removed
         if (mPolicyRegistered != mHandles.size() > 0) {
             doRegister = true;
@@ -642,6 +648,13 @@
             mState = IDLE;
         }
         break;
+    case ACTIVE:
+        for (size_t i = 0; i < mHandles.size(); i++) {
+            if (!mHandles[i]->disconnected()) {
+                mHandles[i]->framesProcessed(mConfig.inputCfg.buffer.frameCount);
+            }
+        }
+        break;
     default: //IDLE , ACTIVE, DESTROYED
         break;
     }
@@ -875,9 +888,9 @@
     // similar to output EFFECT_FLAG_TYPE_INSERT/REPLACE,
     // in which case input channel masks should be used here.
     callback = getCallback();
-    channelMask = callback->channelMask();
+    channelMask = callback->inChannelMask(mId);
     mConfig.inputCfg.channels = channelMask;
-    mConfig.outputCfg.channels = channelMask;
+    mConfig.outputCfg.channels = callback->outChannelMask();
 
     if ((mDescriptor.flags & EFFECT_FLAG_TYPE_MASK) == EFFECT_FLAG_TYPE_AUXILIARY) {
         if (mConfig.inputCfg.channels != AUDIO_CHANNEL_OUT_MONO) {
@@ -1600,7 +1613,7 @@
     return status;
 }
 
-status_t AudioFlinger::EffectModule::setVibratorInfo(const media::AudioVibratorInfo* vibratorInfo)
+status_t AudioFlinger::EffectModule::setVibratorInfo(const media::AudioVibratorInfo& vibratorInfo)
 {
     if (mStatus != NO_ERROR) {
         return mStatus;
@@ -1610,15 +1623,17 @@
         return INVALID_OPERATION;
     }
 
+    const size_t paramCount = 3;
     std::vector<uint8_t> request(
-            sizeof(effect_param_t) + sizeof(int32_t) + 2 * sizeof(float));
+            sizeof(effect_param_t) + sizeof(int32_t) + paramCount * sizeof(float));
     effect_param_t *param = (effect_param_t*) request.data();
     param->psize = sizeof(int32_t);
-    param->vsize = 2 * sizeof(float);
+    param->vsize = paramCount * sizeof(float);
     *(int32_t*)param->data = HG_PARAM_VIBRATOR_INFO;
     float* vibratorInfoPtr = reinterpret_cast<float*>(param->data + sizeof(int32_t));
-    vibratorInfoPtr[0] = vibratorInfo->resonantFrequency;
-    vibratorInfoPtr[1] = vibratorInfo->qFactor;
+    vibratorInfoPtr[0] = vibratorInfo.resonantFrequency;
+    vibratorInfoPtr[1] = vibratorInfo.qFactor;
+    vibratorInfoPtr[2] = vibratorInfo.maxAmplitude;
     std::vector<uint8_t> response;
     status_t status = command(EFFECT_CMD_SET_PARAM, request, sizeof(int32_t), &response);
     if (status == NO_ERROR) {
@@ -1708,10 +1723,11 @@
 AudioFlinger::EffectHandle::EffectHandle(const sp<EffectBase>& effect,
                                          const sp<AudioFlinger::Client>& client,
                                          const sp<media::IEffectClient>& effectClient,
-                                         int32_t priority)
+                                         int32_t priority, bool notifyFramesProcessed)
     : BnEffect(),
     mEffect(effect), mEffectClient(effectClient), mClient(client), mCblk(NULL),
-    mPriority(priority), mHasControl(false), mEnabled(false), mDisconnected(false)
+    mPriority(priority), mHasControl(false), mEnabled(false), mDisconnected(false),
+    mNotifyFramesProcessed(notifyFramesProcessed)
 {
     ALOGV("constructor %p client %p", this, client.get());
 
@@ -2020,6 +2036,13 @@
     }
 }
 
+void AudioFlinger::EffectHandle::framesProcessed(int32_t frames) const
+{
+    if (mEffectClient != 0 && mNotifyFramesProcessed) {
+        mEffectClient->framesProcessed(frames);
+    }
+}
+
 void AudioFlinger::EffectHandle::dumpToBuffer(char* buffer, size_t size)
 {
     bool locked = mCblk != NULL && AudioFlinger::dumpTryLock(mCblk->lock);
@@ -2048,11 +2071,11 @@
       mNewLeftVolume(UINT_MAX), mNewRightVolume(UINT_MAX),
       mEffectCallback(new EffectCallback(wp<EffectChain>(this), thread))
 {
-    mStrategy = AudioSystem::getStrategyForStream(AUDIO_STREAM_MUSIC);
     sp<ThreadBase> p = thread.promote();
     if (p == nullptr) {
         return;
     }
+    mStrategy = p->getStrategyForStream(AUDIO_STREAM_MUSIC);
     mMaxTailBuffers = ((kProcessTailDurationMs * p->sampleRate()) / 1000) /
                                     p->frameCount();
 }
@@ -2125,8 +2148,8 @@
     if (mInBuffer == NULL) {
         return;
     }
-    const size_t frameSize =
-            audio_bytes_per_sample(EFFECT_BUFFER_FORMAT) * mEffectCallback->channelCount();
+    const size_t frameSize = audio_bytes_per_sample(EFFECT_BUFFER_FORMAT)
+            * mEffectCallback->inChannelCount(mEffects[0]->id());
 
     memset(mInBuffer->audioBuffer()->raw, 0, mEffectCallback->frameCount() * frameSize);
     mInBuffer->commit();
@@ -2236,6 +2259,9 @@
                 numSamples * sizeof(int32_t), &halBuffer);
 #endif
         if (result != OK) return result;
+
+        effect->configure();
+
         effect->setInBuffer(halBuffer);
         // auxiliary effects output samples to chain input buffer for further processing
         // by insert effects
@@ -2303,6 +2329,10 @@
             }
         }
 
+        mEffects.insertAt(effect, idx_insert);
+
+        effect->configure();
+
         // always read samples from chain input buffer
         effect->setInBuffer(mInBuffer);
 
@@ -2310,14 +2340,13 @@
         // output buffer, otherwise to chain input buffer
         if (idx_insert == size) {
             if (idx_insert != 0) {
-                mEffects[idx_insert-1]->setOutBuffer(mInBuffer);
                 mEffects[idx_insert-1]->configure();
+                mEffects[idx_insert-1]->setOutBuffer(mInBuffer);
             }
             effect->setOutBuffer(mOutBuffer);
         } else {
             effect->setOutBuffer(mInBuffer);
         }
-        mEffects.insertAt(effect, idx_insert);
 
         ALOGV("addEffect_l() effect %p, added in chain %p at rank %zu", effect.get(), this,
                 idx_insert);
@@ -2350,14 +2379,21 @@
 
             if (type != EFFECT_FLAG_TYPE_AUXILIARY) {
                 if (i == size - 1 && i != 0) {
-                    mEffects[i - 1]->setOutBuffer(mOutBuffer);
                     mEffects[i - 1]->configure();
+                    mEffects[i - 1]->setOutBuffer(mOutBuffer);
                 }
             }
             mEffects.removeAt(i);
+
+            // make sure the input buffer configuration for the new first effect in the chain
+            // is updated if needed (can switch from HAL channel mask to mixer channel mask)
+            if (i == 0 && size > 1) {
+                mEffects[0]->configure();
+                mEffects[0]->setInBuffer(mInBuffer);
+            }
+
             ALOGV("removeEffect_l() effect %p, removed from chain %p at rank %zu", effect.get(),
                     this, i);
-
             break;
         }
     }
@@ -2932,7 +2968,43 @@
     return t->sampleRate();
 }
 
-audio_channel_mask_t AudioFlinger::EffectChain::EffectCallback::channelMask() const {
+audio_channel_mask_t AudioFlinger::EffectChain::EffectCallback::inChannelMask(int id) const {
+    sp<ThreadBase> t = thread().promote();
+    if (t == nullptr) {
+        return AUDIO_CHANNEL_NONE;
+    }
+    sp<EffectChain> c = chain().promote();
+    if (c == nullptr) {
+        return AUDIO_CHANNEL_NONE;
+    }
+
+    if (c->sessionId() != AUDIO_SESSION_OUTPUT_STAGE
+            || c->isFirstEffect(id)) {
+        return t->mixerChannelMask();
+    } else {
+        return t->channelMask();
+    }
+}
+
+uint32_t AudioFlinger::EffectChain::EffectCallback::inChannelCount(int id) const {
+    sp<ThreadBase> t = thread().promote();
+    if (t == nullptr) {
+        return 0;
+    }
+    sp<EffectChain> c = chain().promote();
+    if (c == nullptr) {
+        return 0;
+    }
+
+    if (c->sessionId() != AUDIO_SESSION_OUTPUT_STAGE
+            || c->isFirstEffect(id)) {
+        return audio_channel_count_from_out_mask(t->mixerChannelMask());
+    } else {
+        return t->channelCount();
+    }
+}
+
+audio_channel_mask_t AudioFlinger::EffectChain::EffectCallback::outChannelMask() const {
     sp<ThreadBase> t = thread().promote();
     if (t == nullptr) {
         return AUDIO_CHANNEL_NONE;
@@ -2940,7 +3012,7 @@
     return t->channelMask();
 }
 
-uint32_t AudioFlinger::EffectChain::EffectCallback::channelCount() const {
+uint32_t AudioFlinger::EffectChain::EffectCallback::outChannelCount() const {
     sp<ThreadBase> t = thread().promote();
     if (t == nullptr) {
         return 0;
@@ -3143,7 +3215,8 @@
         } else {
             mHalEffect->setDevices({mDevice});
         }
-        *handle = new EffectHandle(mHalEffect, nullptr, nullptr, 0 /*priority*/);
+        *handle = new EffectHandle(mHalEffect, nullptr, nullptr, 0 /*priority*/,
+                                   mNotifyFramesProcessed);
         status = (*handle)->initCheck();
         if (status == OK) {
             status = mHalEffect->addHandle((*handle).get());
@@ -3169,7 +3242,8 @@
         int enabled;
         *handle = thread->createEffect_l(nullptr, nullptr, 0, AUDIO_SESSION_DEVICE,
                                          const_cast<effect_descriptor_t *>(&mDescriptor),
-                                         &enabled, &status, false, false /*probe*/);
+                                         &enabled, &status, false, false /*probe*/,
+                                         mNotifyFramesProcessed);
         ALOGV("%s thread->createEffect_l status %d", __func__, status);
     } else {
         status = BAD_VALUE;
@@ -3364,7 +3438,8 @@
     return proxy->sampleRate();
 }
 
-audio_channel_mask_t AudioFlinger::DeviceEffectProxy::ProxyCallback::channelMask() const {
+audio_channel_mask_t AudioFlinger::DeviceEffectProxy::ProxyCallback::inChannelMask(
+        int id __unused) const {
     sp<DeviceEffectProxy> proxy = mProxy.promote();
     if (proxy == nullptr) {
         return AUDIO_CHANNEL_OUT_STEREO;
@@ -3372,7 +3447,23 @@
     return proxy->channelMask();
 }
 
-uint32_t AudioFlinger::DeviceEffectProxy::ProxyCallback::channelCount() const {
+uint32_t AudioFlinger::DeviceEffectProxy::ProxyCallback::inChannelCount(int id __unused) const {
+    sp<DeviceEffectProxy> proxy = mProxy.promote();
+    if (proxy == nullptr) {
+        return 2;
+    }
+    return proxy->channelCount();
+}
+
+audio_channel_mask_t AudioFlinger::DeviceEffectProxy::ProxyCallback::outChannelMask() const {
+    sp<DeviceEffectProxy> proxy = mProxy.promote();
+    if (proxy == nullptr) {
+        return AUDIO_CHANNEL_OUT_STEREO;
+    }
+    return proxy->channelMask();
+}
+
+uint32_t AudioFlinger::DeviceEffectProxy::ProxyCallback::outChannelCount() const {
     sp<DeviceEffectProxy> proxy = mProxy.promote();
     if (proxy == nullptr) {
         return 2;
diff --git a/services/audioflinger/Effects.h b/services/audioflinger/Effects.h
index a727e04..389ff7b 100644
--- a/services/audioflinger/Effects.h
+++ b/services/audioflinger/Effects.h
@@ -34,8 +34,10 @@
     virtual bool isOffloadOrDirect() const = 0;
     virtual bool isOffloadOrMmap() const = 0;
     virtual uint32_t sampleRate() const = 0;
-    virtual audio_channel_mask_t channelMask() const = 0;
-    virtual uint32_t channelCount() const = 0;
+    virtual audio_channel_mask_t inChannelMask(int id) const = 0;
+    virtual uint32_t inChannelCount(int id) const = 0;
+    virtual audio_channel_mask_t outChannelMask() const = 0;
+    virtual uint32_t outChannelCount() const = 0;
     virtual audio_channel_mask_t hapticChannelMask() const = 0;
     virtual size_t frameCount() const = 0;
 
@@ -64,6 +66,8 @@
     virtual void resetVolume() = 0;
 
     virtual wp<EffectChain> chain() const = 0;
+
+    virtual bool isAudioPolicyReady() const = 0;
 };
 
 // EffectBase(EffectModule) and EffectChain classes both have their own mutex to protect
@@ -164,6 +168,16 @@
 
     void             dump(int fd, const Vector<String16>& args);
 
+protected:
+    bool             isInternal_l() const {
+                         for (auto handle : mHandles) {
+                            if (handle->client() != nullptr) {
+                                return false;
+                            }
+                         }
+                         return true;
+                     }
+
 private:
     friend class AudioFlinger;      // for mHandles
     bool             mPinned = false;
@@ -259,7 +273,7 @@
     bool             isHapticGenerator() const;
 
     status_t         setHapticIntensity(int id, int intensity);
-    status_t         setVibratorInfo(const media::AudioVibratorInfo* vibratorInfo);
+    status_t         setVibratorInfo(const media::AudioVibratorInfo& vibratorInfo);
 
     void             dump(int fd, const Vector<String16>& args);
 
@@ -327,7 +341,7 @@
     EffectHandle(const sp<EffectBase>& effect,
             const sp<AudioFlinger::Client>& client,
             const sp<media::IEffectClient>& effectClient,
-            int32_t priority);
+            int32_t priority, bool notifyFramesProcessed);
     virtual ~EffectHandle();
     virtual status_t initCheck();
 
@@ -342,6 +356,8 @@
     android::binder::Status disconnect() override;
     android::binder::Status getCblk(media::SharedFileRegion* _aidl_return) override;
 
+    sp<Client> client() const { return mClient; }
+
 private:
     void disconnect(bool unpinIfLast);
 
@@ -356,6 +372,8 @@
     void setEnabled(bool enabled);
     bool enabled() const { return mEnabled; }
 
+    void framesProcessed(int32_t frames) const;
+
     // Getters
     wp<EffectBase> effect() const { return mEffect; }
     int id() const {
@@ -389,6 +407,8 @@
     bool mEnabled;                           // cached enable state: needed when the effect is
                                              // restored after being suspended
     bool mDisconnected;                      // Set to true by disconnect()
+    const bool mNotifyFramesProcessed;       // true if the client callback event
+                                             // EVENT_FRAMES_PROCESSED must be generated
 };
 
 // the EffectChain class represents a group of effects associated to one audio session.
@@ -511,6 +531,8 @@
     sp<EffectCallbackInterface> effectCallback() const { return mEffectCallback; }
     wp<ThreadBase> thread() const { return mEffectCallback->thread(); }
 
+    bool isFirstEffect(int id) const { return !mEffects.isEmpty() && id == mEffects[0]->id(); }
+
     void dump(int fd, const Vector<String16>& args);
 
 private:
@@ -544,8 +566,10 @@
         bool isOffloadOrMmap() const override;
 
         uint32_t sampleRate() const override;
-        audio_channel_mask_t channelMask() const override;
-        uint32_t channelCount() const override;
+        audio_channel_mask_t inChannelMask(int id) const override;
+        uint32_t inChannelCount(int id) const override;
+        audio_channel_mask_t outChannelMask() const override;
+        uint32_t outChannelCount() const override;
         audio_channel_mask_t hapticChannelMask() const override;
         size_t frameCount() const override;
         uint32_t latency() const override;
@@ -566,6 +590,10 @@
 
         wp<EffectChain> chain() const override { return mChain; }
 
+        bool isAudioPolicyReady() const override {
+            return mAudioFlinger.isAudioPolicyReady();
+        }
+
         wp<ThreadBase> thread() const { return mThread.load(); }
 
         void setThread(const wp<ThreadBase>& thread) {
@@ -643,11 +671,11 @@
 public:
         DeviceEffectProxy (const AudioDeviceTypeAddr& device,
                 const sp<DeviceEffectManagerCallback>& callback,
-                effect_descriptor_t *desc, int id)
+                effect_descriptor_t *desc, int id, bool notifyFramesProcessed)
             : EffectBase(callback, desc, id, AUDIO_SESSION_DEVICE, false),
                 mDevice(device), mManagerCallback(callback),
-                mMyCallback(new ProxyCallback(wp<DeviceEffectProxy>(this),
-                                              callback)) {}
+                mMyCallback(new ProxyCallback(wp<DeviceEffectProxy>(this), callback)),
+                mNotifyFramesProcessed(notifyFramesProcessed) {}
 
     status_t setEnabled(bool enabled, bool fromHandle) override;
     sp<DeviceEffectProxy> asDeviceEffectProxy() override { return this; }
@@ -694,8 +722,10 @@
         bool isOffloadOrMmap() const override { return false; }
 
         uint32_t sampleRate() const override;
-        audio_channel_mask_t channelMask() const override;
-        uint32_t channelCount() const override;
+        audio_channel_mask_t inChannelMask(int id) const override;
+        uint32_t inChannelCount(int id) const override;
+        audio_channel_mask_t outChannelMask() const override;
+        uint32_t outChannelCount() const override;
         audio_channel_mask_t hapticChannelMask() const override { return AUDIO_CHANNEL_NONE; }
         size_t frameCount() const override  { return 0; }
         uint32_t latency() const override  { return 0; }
@@ -716,6 +746,10 @@
 
         wp<EffectChain> chain() const override { return nullptr; }
 
+        bool isAudioPolicyReady() const override {
+            return mManagerCallback->isAudioPolicyReady();
+        }
+
         int newEffectId();
 
     private:
@@ -734,4 +768,5 @@
     std::map<audio_patch_handle_t, sp<EffectHandle>> mEffectHandles; // protected by mProxyLock
     sp<EffectModule> mHalEffect; // protected by mProxyLock
     struct audio_port_config mDevicePort = { .id = AUDIO_PORT_HANDLE_NONE };
+    const bool mNotifyFramesProcessed;
 };
diff --git a/services/audioflinger/FastMixer.cpp b/services/audioflinger/FastMixer.cpp
index 88d4eaf..fc34d95 100644
--- a/services/audioflinger/FastMixer.cpp
+++ b/services/audioflinger/FastMixer.cpp
@@ -204,6 +204,8 @@
                 (void *)(uintptr_t)fastTrack->mHapticPlaybackEnabled);
         mMixer->setParameter(index, AudioMixer::TRACK, AudioMixer::HAPTIC_INTENSITY,
                 (void *)(uintptr_t)fastTrack->mHapticIntensity);
+        mMixer->setParameter(index, AudioMixer::TRACK, AudioMixer::HAPTIC_MAX_AMPLITUDE,
+                (void *)(&(fastTrack->mHapticMaxAmplitude)));
 
         mMixer->enable(index);
         break;
diff --git a/services/audioflinger/FastMixerState.h b/services/audioflinger/FastMixerState.h
index 857d3de..ce3cc14 100644
--- a/services/audioflinger/FastMixerState.h
+++ b/services/audioflinger/FastMixerState.h
@@ -17,6 +17,8 @@
 #ifndef ANDROID_AUDIO_FAST_MIXER_STATE_H
 #define ANDROID_AUDIO_FAST_MIXER_STATE_H
 
+#include <math.h>
+
 #include <audio_utils/minifloat.h>
 #include <system/audio.h>
 #include <media/AudioMixer.h>
@@ -51,6 +53,7 @@
     int                     mGeneration;     // increment when any field is assigned
     bool                    mHapticPlaybackEnabled = false; // haptic playback is enabled or not
     os::HapticScale         mHapticIntensity = os::HapticScale::MUTE; // intensity of haptic data
+    float                   mHapticMaxAmplitude = NAN; // max amplitude allowed for haptic data
 };
 
 // Represents a single state of the fast mixer
diff --git a/services/audioflinger/PatchPanel.cpp b/services/audioflinger/PatchPanel.cpp
index a381c7d..93118b8 100644
--- a/services/audioflinger/PatchPanel.cpp
+++ b/services/audioflinger/PatchPanel.cpp
@@ -258,6 +258,7 @@
                             reinterpret_cast<PlaybackThread*>(thread.get()), false /*closeThread*/);
                 } else {
                     audio_config_t config = AUDIO_CONFIG_INITIALIZER;
+                    audio_config_base_t mixerConfig = AUDIO_CONFIG_BASE_INITIALIZER;
                     audio_io_handle_t output = AUDIO_IO_HANDLE_NONE;
                     audio_output_flags_t flags = AUDIO_OUTPUT_FLAG_NONE;
                     if (patch->sinks[0].config_mask & AUDIO_PORT_CONFIG_SAMPLE_RATE) {
@@ -276,6 +277,7 @@
                                                             patch->sinks[0].ext.device.hw_module,
                                                             &output,
                                                             &config,
+                                                            &mixerConfig,
                                                             outputDevice,
                                                             outputDeviceAddress,
                                                             flags);
diff --git a/services/audioflinger/PlaybackTracks.h b/services/audioflinger/PlaybackTracks.h
index 0929055..b069462 100644
--- a/services/audioflinger/PlaybackTracks.h
+++ b/services/audioflinger/PlaybackTracks.h
@@ -19,6 +19,8 @@
     #error This header file should only be included from AudioFlinger.h
 #endif
 
+#include <math.h>
+
 // Checks and monitors OP_PLAY_AUDIO
 class OpPlayAudioMonitor : public RefBase {
 public:
@@ -161,6 +163,8 @@
             }
             /** Return at what intensity to play haptics, used in mixer. */
             os::HapticScale getHapticIntensity() const { return mHapticIntensity; }
+            /** Return the maximum amplitude allowed for haptics data, used in mixer. */
+            float getHapticMaxAmplitude() const { return mHapticMaxAmplitude; }
             /** Set intensity of haptic playback, should be set after querying vibrator service. */
             void    setHapticIntensity(os::HapticScale hapticIntensity) {
                 if (os::isValidHapticScale(hapticIntensity)) {
@@ -168,6 +172,12 @@
                     setHapticPlaybackEnabled(mHapticIntensity != os::HapticScale::MUTE);
                 }
             }
+            /** Set maximum amplitude allowed for haptic data, should be set after querying
+             *  vibrator service.
+             */
+            void    setHapticMaxAmplitude(float maxAmplitude) {
+                mHapticMaxAmplitude = maxAmplitude;
+            }
             sp<os::ExternalVibration> getExternalVibration() const { return mExternalVibration; }
 
             void    setTeePatches(TeePatches teePatches);
@@ -185,6 +195,10 @@
 
     audio_output_flags_t getOutputFlags() const { return mFlags; }
     float getSpeed() const { return mSpeed; }
+
+    bool canBeSpatialized() const { return (mAttr.flags
+            & (AUDIO_FLAG_CONTENT_SPATIALIZED | AUDIO_FLAG_NEVER_SPATIALIZE)) == 0; }
+
 protected:
     // for numerous
     friend class PlaybackThread;
@@ -282,6 +296,8 @@
     bool                mHapticPlaybackEnabled = false; // indicates haptic playback enabled or not
     // intensity to play haptic data
     os::HapticScale mHapticIntensity = os::HapticScale::MUTE;
+    // max amplitude allowed for haptic data
+    float mHapticMaxAmplitude = NAN;
     class AudioVibrationController : public os::BnExternalVibrationController {
     public:
         explicit AudioVibrationController(Track* track) : mTrack(track) {}
diff --git a/services/audioflinger/Threads.cpp b/services/audioflinger/Threads.cpp
index b9cdab8..fddd3de 100644
--- a/services/audioflinger/Threads.cpp
+++ b/services/audioflinger/Threads.cpp
@@ -50,8 +50,10 @@
 #include <audio_utils/format.h>
 #include <audio_utils/minifloat.h>
 #include <audio_utils/safe_math.h>
-#include <system/audio_effects/effect_ns.h>
 #include <system/audio_effects/effect_aec.h>
+#include <system/audio_effects/effect_downmix.h>
+#include <system/audio_effects/effect_ns.h>
+#include <system/audio_effects/effect_spatializer.h>
 #include <system/audio.h>
 
 // NBAIO implementations
@@ -507,6 +509,8 @@
         return "MMAP_PLAYBACK";
     case MMAP_CAPTURE:
         return "MMAP_CAPTURE";
+    case SPATIALIZER:
+        return "SPATIALIZER";
     default:
         return "unknown";
     }
@@ -722,6 +726,19 @@
     sendConfigEvent_l(configEvent);
 }
 
+void AudioFlinger::ThreadBase::sendCheckOutputStageEffectsEvent()
+{
+    Mutex::Autolock _l(mLock);
+    sendCheckOutputStageEffectsEvent_l();
+}
+
+void AudioFlinger::ThreadBase::sendCheckOutputStageEffectsEvent_l()
+{
+    sp<ConfigEvent> configEvent =
+            (ConfigEvent *)new CheckOutputStageEffectsEvent();
+    sendConfigEvent_l(configEvent);
+}
+
 // post condition: mConfigEvents.isEmpty()
 void AudioFlinger::ThreadBase::processConfigEvents_l()
 {
@@ -784,6 +801,11 @@
                     (ResizeBufferConfigEventData *)event->mData.get();
             resizeInputBuffer_l(data->mMaxSharedAudioHistoryMs);
         } break;
+
+        case CFG_EVENT_CHECK_OUTPUT_STAGE_EFFECTS: {
+            setCheckOutputStageEffects();
+        } break;
+
         default:
             ALOG_ASSERT(false, "processConfigEvents_l() unknown event type %d", event->mType);
             break;
@@ -1008,6 +1030,8 @@
         return String16("MmapPlayback");
     case MMAP_CAPTURE:
         return String16("MmapCapture");
+    case SPATIALIZER:
+        return String16("AudioSpatial");
     default:
         ALOG_ASSERT(false);
         return String16("AudioUnknown");
@@ -1312,6 +1336,13 @@
         return BAD_VALUE;
     }
 
+    if (memcmp(&desc->type, FX_IID_SPATIALIZER, sizeof(effect_uuid_t)) == 0
+            && mType != SPATIALIZER) {
+        ALOGW("%s: attempt to create a spatializer effect on a thread of type %d",
+                __func__, mType);
+        return BAD_VALUE;
+    }
+
     switch (mType) {
     case MIXER: {
 #ifndef MULTICHANNEL_EFFECT_CHAIN
@@ -1401,6 +1432,13 @@
             return BAD_VALUE;
         }
         break;
+    case SPATIALIZER:
+        if (!audio_is_global_session(sessionId)) {
+            ALOGW("checkEffectCompatibility_l(): non global effect %s on SPATIALIZER"
+                    " thread %s", desc->name, mThreadName);
+            return BAD_VALUE;
+        }
+        break;
     default:
         LOG_ALWAYS_FATAL("checkEffectCompatibility_l(): wrong thread type %d", mType);
     }
@@ -1418,7 +1456,8 @@
         int *enabled,
         status_t *status,
         bool pinned,
-        bool probe)
+        bool probe,
+        bool notifyFramesProcessed)
 {
     sp<EffectModule> effect;
     sp<EffectHandle> handle;
@@ -1477,18 +1516,19 @@
         if (effect->isHapticGenerator()) {
             // TODO(b/184194057): Use the vibrator information from the vibrator that will be used
             // for the HapticGenerator.
-            const media::AudioVibratorInfo* defaultVibratorInfo =
-                    mAudioFlinger->getDefaultVibratorInfo_l();
-            if (defaultVibratorInfo != nullptr) {
+            const std::optional<media::AudioVibratorInfo> defaultVibratorInfo =
+                    std::move(mAudioFlinger->getDefaultVibratorInfo_l());
+            if (defaultVibratorInfo) {
                 // Only set the vibrator info when it is a valid one.
-                effect->setVibratorInfo(defaultVibratorInfo);
+                effect->setVibratorInfo(*defaultVibratorInfo);
             }
         }
         // create effect handle and connect it to effect module
-        handle = new EffectHandle(effect, client, effectClient, priority);
+        handle = new EffectHandle(effect, client, effectClient, priority, notifyFramesProcessed);
         lStatus = handle->initCheck();
         if (lStatus == OK) {
             lStatus = effect->addHandle(handle.get());
+            sendCheckOutputStageEffectsEvent_l();
         }
         if (enabled != NULL) {
             *enabled = (int)effect->isEnabled();
@@ -1531,6 +1571,7 @@
         if (remove) {
             removeEffect_l(effect, true);
         }
+        sendCheckOutputStageEffectsEvent_l();
     }
     if (remove) {
         mAudioFlinger->updateOrphanEffectChains(effect);
@@ -1888,6 +1929,14 @@
     item->selfrecord();
 }
 
+product_strategy_t AudioFlinger::ThreadBase::getStrategyForStream(audio_stream_type_t stream) const
+{
+    if (!mAudioFlinger->isAudioPolicyReady()) {
+        return PRODUCT_STRATEGY_NONE;
+    }
+    return AudioSystem::getStrategyForStream(stream);
+}
+
 // ----------------------------------------------------------------------------
 //      Playback
 // ----------------------------------------------------------------------------
@@ -1896,15 +1945,16 @@
                                              AudioStreamOut* output,
                                              audio_io_handle_t id,
                                              type_t type,
-                                             bool systemReady)
+                                             bool systemReady,
+                                             audio_config_base_t *mixerConfig)
     :   ThreadBase(audioFlinger, id, type, systemReady, true /* isOut */),
         mNormalFrameCount(0), mSinkBuffer(NULL),
-        mMixerBufferEnabled(AudioFlinger::kEnableExtendedPrecision),
+        mMixerBufferEnabled(AudioFlinger::kEnableExtendedPrecision || type == SPATIALIZER),
         mMixerBuffer(NULL),
         mMixerBufferSize(0),
         mMixerBufferFormat(AUDIO_FORMAT_INVALID),
         mMixerBufferValid(false),
-        mEffectBufferEnabled(AudioFlinger::kEnableExtendedPrecision),
+        mEffectBufferEnabled(AudioFlinger::kEnableExtendedPrecision || type == SPATIALIZER),
         mEffectBuffer(NULL),
         mEffectBufferSize(0),
         mEffectBufferFormat(AUDIO_FORMAT_INVALID),
@@ -1956,8 +2006,18 @@
                 mOutput->audioHwDev->moduleName(), AUDIO_HARDWARE_MODULE_ID_MSD) == 0;
     }
 
+    if (mixerConfig != nullptr && mixerConfig->channel_mask != AUDIO_CHANNEL_NONE) {
+        mMixerChannelMask = mixerConfig->channel_mask;
+    }
+
     readOutputParameters_l();
 
+    if (mType != SPATIALIZER
+            && mMixerChannelMask != mChannelMask) {
+        LOG_ALWAYS_FATAL("HAL channel mask %#x does not match mixer channel mask %#x",
+                mChannelMask, mMixerChannelMask);
+    }
+
     // TODO: We may also match on address as well as device type for
     // AUDIO_DEVICE_OUT_BUS, AUDIO_DEVICE_OUT_ALL_A2DP, AUDIO_DEVICE_OUT_REMOTE_SUBMIX
     if (type == MIXER || type == DIRECT || type == OFFLOAD) {
@@ -1986,6 +2046,7 @@
     free(mSinkBuffer);
     free(mMixerBuffer);
     free(mEffectBuffer);
+    free(mEffectToSinkBuffer);
 }
 
 // Thread virtuals
@@ -2080,10 +2141,12 @@
     write(fd, result.string(), result.size());
 }
 
-void AudioFlinger::PlaybackThread::dumpInternals_l(int fd, const Vector<String16>& args __unused)
+void AudioFlinger::PlaybackThread::dumpInternals_l(int fd, const Vector<String16>& args)
 {
     dprintf(fd, "  Master volume: %f\n", mMasterVolume);
     dprintf(fd, "  Master mute: %s\n", mMasterMute ? "on" : "off");
+    dprintf(fd, "  Mixer channel Mask: %#x (%s)\n",
+            mMixerChannelMask, channelMaskToString(mMixerChannelMask, true /* output */).c_str());
     if (mHapticChannelMask != AUDIO_CHANNEL_NONE) {
         dprintf(fd, "  Haptic channel mask: %#x (%s)\n", mHapticChannelMask,
                 channelMaskToString(mHapticChannelMask, true /* output */).c_str());
@@ -2109,7 +2172,7 @@
     }
     if (output != nullptr) {
         dprintf(fd, "  Hal stream dump:\n");
-        (void)output->stream->dump(fd);
+        (void)output->stream->dump(fd, args);
     }
 }
 
@@ -2397,11 +2460,11 @@
         // all tracks in same audio session must share the same routing strategy otherwise
         // conflicts will happen when tracks are moved from one output to another by audio policy
         // manager
-        product_strategy_t strategy = AudioSystem::getStrategyForStream(streamType);
+        product_strategy_t strategy = getStrategyForStream(streamType);
         for (size_t i = 0; i < mTracks.size(); ++i) {
             sp<Track> t = mTracks[i];
             if (t != 0 && t->isExternalTrack()) {
-                product_strategy_t actual = AudioSystem::getStrategyForStream(t->streamType());
+                product_strategy_t actual = getStrategyForStream(t->streamType());
                 if (sessionId == t->sessionId() && strategy != actual) {
                     ALOGE("createTrack_l() mismatched strategy; expected %u but found %u",
                             strategy, actual);
@@ -2445,7 +2508,7 @@
         if (chain != 0) {
             ALOGV("createTrack_l() setting main buffer %p", chain->inBuffer());
             track->setMainBuffer(chain->inBuffer());
-            chain->setStrategy(AudioSystem::getStrategyForStream(track->streamType()));
+            chain->setStrategy(getStrategyForStream(track->streamType()));
             chain->incTrackCnt();
         }
 
@@ -2613,8 +2676,19 @@
             mLock.unlock();
             const int intensity = AudioFlinger::onExternalVibrationStart(
                     track->getExternalVibration());
+            std::optional<media::AudioVibratorInfo> vibratorInfo;
+            {
+                // TODO(b/184194780): Use the vibrator information from the vibrator that will be
+                // used to play this track.
+                Mutex::Autolock _l(mAudioFlinger->mLock);
+                vibratorInfo = std::move(mAudioFlinger->getDefaultVibratorInfo_l());
+            }
             mLock.lock();
             track->setHapticIntensity(static_cast<os::HapticScale>(intensity));
+            if (vibratorInfo) {
+                track->setHapticMaxAmplitude(vibratorInfo->maxAmplitude);
+            }
+
             // Haptic playback should be enabled by vibrator service.
             if (track->getHapticPlaybackEnabled()) {
                 // Disable haptic playback of all active track to ensure only
@@ -2814,14 +2888,20 @@
     if (!audio_is_output_channel(mChannelMask)) {
         LOG_ALWAYS_FATAL("HAL channel mask %#x not valid for output", mChannelMask);
     }
-    if ((mType == MIXER || mType == DUPLICATING)
-            && !isValidPcmSinkChannelMask(mChannelMask)) {
+    if (hasMixer() && !isValidPcmSinkChannelMask(mChannelMask)) {
         LOG_ALWAYS_FATAL("HAL channel mask %#x not supported for mixed output",
                 mChannelMask);
     }
+
+    if (mMixerChannelMask == AUDIO_CHANNEL_NONE) {
+        mMixerChannelMask = mChannelMask;
+    }
+
     mChannelCount = audio_channel_count_from_out_mask(mChannelMask);
     mBalance.setChannelMask(mChannelMask);
 
+    uint32_t mixerChannelCount = audio_channel_count_from_out_mask(mMixerChannelMask);
+
     // Get actual HAL format.
     status_t result = mOutput->stream->getAudioProperties(nullptr, nullptr, &mHALFormat);
     LOG_ALWAYS_FATAL_IF(result != OK, "Error when retrieving output stream format: %d", result);
@@ -2831,8 +2911,7 @@
     if (!audio_is_valid_format(mFormat)) {
         LOG_ALWAYS_FATAL("HAL format %#x not valid for output", mFormat);
     }
-    if ((mType == MIXER || mType == DUPLICATING)
-            && !isValidPcmSinkFormat(mFormat)) {
+    if (hasMixer() && !isValidPcmSinkFormat(mFormat)) {
         LOG_FATAL("HAL format %#x not supported for mixed output",
                 mFormat);
     }
@@ -2841,7 +2920,7 @@
     LOG_ALWAYS_FATAL_IF(result != OK,
             "Error when retrieving output stream buffer size: %d", result);
     mFrameCount = mBufferSize / mFrameSize;
-    if ((mType == MIXER || mType == DUPLICATING) && (mFrameCount & 15)) {
+    if (hasMixer() && (mFrameCount & 15)) {
         ALOGW("HAL output buffer size is %zu frames but AudioMixer requires multiples of 16 frames",
                 mFrameCount);
     }
@@ -2914,7 +2993,7 @@
     }
     mNormalFrameCount = multiplier * mFrameCount;
     // round up to nearest 16 frames to satisfy AudioMixer
-    if (mType == MIXER || mType == DUPLICATING) {
+    if (hasMixer()) {
         mNormalFrameCount = (mNormalFrameCount + 15) & ~15;
     }
     ALOGI("HAL output buffer size %zu frames, normal sink buffer size %zu frames", mFrameCount,
@@ -2930,18 +3009,25 @@
     // Originally this was int16_t[] array, need to remove legacy implications.
     free(mSinkBuffer);
     mSinkBuffer = NULL;
+    free(mEffectToSinkBuffer);
+    mEffectToSinkBuffer = nullptr;
+
     // For sink buffer size, we use the frame size from the downstream sink to avoid problems
     // with non PCM formats for compressed music, e.g. AAC, and Offload threads.
     const size_t sinkBufferSize = mNormalFrameCount * mFrameSize;
     (void)posix_memalign(&mSinkBuffer, 32, sinkBufferSize);
 
+    if (mType == SPATIALIZER) {
+        (void)posix_memalign(&mEffectToSinkBuffer, 32, sinkBufferSize);
+    }
+
     // We resize the mMixerBuffer according to the requirements of the sink buffer which
     // drives the output.
     free(mMixerBuffer);
     mMixerBuffer = NULL;
     if (mMixerBufferEnabled) {
         mMixerBufferFormat = AUDIO_FORMAT_PCM_FLOAT; // no longer valid: AUDIO_FORMAT_PCM_16_BIT.
-        mMixerBufferSize = mNormalFrameCount * mChannelCount
+        mMixerBufferSize = mNormalFrameCount * mixerChannelCount
                 * audio_bytes_per_sample(mMixerBufferFormat);
         (void)posix_memalign(&mMixerBuffer, 32, mMixerBufferSize);
     }
@@ -2949,7 +3035,7 @@
     mEffectBuffer = NULL;
     if (mEffectBufferEnabled) {
         mEffectBufferFormat = EFFECT_BUFFER_FORMAT;
-        mEffectBufferSize = mNormalFrameCount * mChannelCount
+        mEffectBufferSize = mNormalFrameCount * mixerChannelCount
                 * audio_bytes_per_sample(mEffectBufferFormat);
         (void)posix_memalign(&mEffectBuffer, 32, mEffectBufferSize);
     }
@@ -2958,6 +3044,7 @@
     mChannelMask = static_cast<audio_channel_mask_t>(mChannelMask & ~mHapticChannelMask);
     mHapticChannelCount = audio_channel_count_from_out_mask(mHapticChannelMask);
     mChannelCount -= mHapticChannelCount;
+    mMixerChannelMask = static_cast<audio_channel_mask_t>(mMixerChannelMask & ~mHapticChannelMask);
 
     // force reconfiguration of effect chains and engines to take new buffer size and audio
     // parameters into account
@@ -3051,15 +3138,15 @@
     // session AUDIO_SESSION_OUTPUT_MIX is placed in same strategy as MUSIC stream so that
     // it is moved to correct output by audio policy manager when A2DP is connected or disconnected
     if (sessionId == AUDIO_SESSION_OUTPUT_MIX) {
-        return AudioSystem::getStrategyForStream(AUDIO_STREAM_MUSIC);
+        return getStrategyForStream(AUDIO_STREAM_MUSIC);
     }
     for (size_t i = 0; i < mTracks.size(); i++) {
         sp<Track> track = mTracks[i];
         if (sessionId == track->sessionId() && !track->isInvalid()) {
-            return AudioSystem::getStrategyForStream(track->streamType());
+            return getStrategyForStream(track->streamType());
         }
     }
-    return AudioSystem::getStrategyForStream(AUDIO_STREAM_MUSIC);
+    return getStrategyForStream(AUDIO_STREAM_MUSIC);
 }
 
 
@@ -3348,7 +3435,8 @@
         // Only one effect chain can be present in direct output thread and it uses
         // the sink buffer as input
         if (mType != DIRECT) {
-            size_t numSamples = mNormalFrameCount * (mChannelCount + mHapticChannelCount);
+            size_t numSamples = mNormalFrameCount
+                    * (audio_channel_count_from_out_mask(mMixerChannelMask) + mHapticChannelCount);
             status_t result = mAudioFlinger->mEffectsFactoryHal->allocateBuffer(
                     numSamples * sizeof(effect_buffer_t),
                     &halInBuffer);
@@ -3531,6 +3619,8 @@
 
     audio_patch_handle_t lastDownstreamPatchHandle = AUDIO_PATCH_HANDLE_NONE;
 
+    sendCheckOutputStageEffectsEvent();
+
     // loopCount is used for statistics and diagnostics.
     for (int64_t loopCount = 0; !exitPending(); ++loopCount)
     {
@@ -3587,11 +3677,18 @@
             }
         }
 
+        if (mCheckOutputStageEffects.exchange(false)) {
+            checkOutputStageEffects();
+        }
+
         { // scope for mLock
 
             Mutex::Autolock _l(mLock);
 
             processConfigEvents_l();
+            if (mCheckOutputStageEffects.load()) {
+                continue;
+            }
 
             // See comment at declaration of logString for why this is done under mLock
             if (logString != NULL) {
@@ -3754,6 +3851,8 @@
             //
             // mMixerBufferValid is only set true by MixerThread::prepareTracks_l().
             // TODO use mSleepTimeUs == 0 as an additional condition.
+            uint32_t mixerChannelCount = mEffectBufferValid ?
+                        audio_channel_count_from_out_mask(mMixerChannelMask) : mChannelCount;
             if (mMixerBufferValid) {
                 void *buffer = mEffectBufferValid ? mEffectBuffer : mSinkBuffer;
                 audio_format_t format = mEffectBufferValid ? mEffectBufferFormat : mFormat;
@@ -3774,7 +3873,7 @@
                 }
 
                 memcpy_by_audio_format(buffer, format, mMixerBuffer, mMixerBufferFormat,
-                        mNormalFrameCount * (mChannelCount + mHapticChannelCount));
+                        mNormalFrameCount * (mixerChannelCount + mHapticChannelCount));
 
                 // If we're going directly to the sink and there are haptic channels,
                 // we should adjust channels as the sample data is partially interleaved
@@ -3807,8 +3906,11 @@
                             && activeHapticSessionId == effectChains[i]->sessionId()) {
                         // Haptic data is active in this case, copy it directly from
                         // in buffer to out buffer.
+                        uint32_t channelCount =
+                                effectChains[i]->sessionId() == AUDIO_SESSION_OUTPUT_STAGE ?
+                                        mixerChannelCount : mChannelCount;
                         const size_t audioBufferSize = mNormalFrameCount
-                                * audio_bytes_per_frame(mChannelCount, EFFECT_BUFFER_FORMAT);
+                                * audio_bytes_per_frame(channelCount, EFFECT_BUFFER_FORMAT);
                         memcpy_by_audio_format(
                                 (uint8_t*)effectChains[i]->outBuffer() + audioBufferSize,
                                 EFFECT_BUFFER_FORMAT,
@@ -3848,8 +3950,23 @@
                 mBalance.process((float *)mEffectBuffer, mNormalFrameCount);
             }
 
-            memcpy_by_audio_format(mSinkBuffer, mFormat, mEffectBuffer, mEffectBufferFormat,
-                    mNormalFrameCount * (mChannelCount + mHapticChannelCount));
+            if (mType == SPATIALIZER) {
+                memcpy_by_audio_format(mEffectToSinkBuffer, mFormat, mEffectBuffer,
+                        mEffectBufferFormat,
+                        mNormalFrameCount * (mChannelCount + mHapticChannelCount));
+                accumulate_by_audio_format(mSinkBuffer, mEffectToSinkBuffer, mFormat,
+                                           mNormalFrameCount * mChannelCount);
+                const size_t audioBufferSize = mNormalFrameCount
+                        * audio_bytes_per_frame(mChannelCount, mFormat);
+                memcpy_by_audio_format(
+                        (uint8_t*)mSinkBuffer + audioBufferSize,
+                        mFormat,
+                        (uint8_t*)mEffectToSinkBuffer + audioBufferSize,
+                        mFormat, mNormalFrameCount * mHapticChannelCount);
+            } else {
+                memcpy_by_audio_format(mSinkBuffer, mFormat, mEffectBuffer, mEffectBufferFormat,
+                        mNormalFrameCount * (mChannelCount + mHapticChannelCount));
+            }
             // The sample data is partially interleaved when haptic channels exist,
             // we need to adjust channels here.
             if (mHapticChannelCount > 0) {
@@ -4448,8 +4565,8 @@
 // ----------------------------------------------------------------------------
 
 AudioFlinger::MixerThread::MixerThread(const sp<AudioFlinger>& audioFlinger, AudioStreamOut* output,
-        audio_io_handle_t id, bool systemReady, type_t type)
-    :   PlaybackThread(audioFlinger, output, id, type, systemReady),
+        audio_io_handle_t id, bool systemReady, type_t type, audio_config_base_t *mixerConfig)
+    :   PlaybackThread(audioFlinger, output, id, type, systemReady, mixerConfig),
         // mAudioMixer below
         // mFastMixer below
         mFastMixerFutex(0),
@@ -4504,6 +4621,7 @@
                 && Intersection(outDeviceTypes(), getAudioDeviceOutAllA2dpSet()).empty();
         break;
     }
+    ALOG_ASSERT(initFastMixer && mType == SPATIALIZER);
     ALOGW_IF(initFastMixer == false && mFrameCount < mNormalFrameCount,
             "FastMixer is preferred for this sink as frameCount %zu is less than threshold %zu",
             mFrameCount, mNormalFrameCount);
@@ -4566,6 +4684,7 @@
         fastTrack->mFormat = mFormat; // mPipeSink format for audio to FastMixer
         fastTrack->mHapticPlaybackEnabled = mHapticChannelMask != AUDIO_CHANNEL_NONE;
         fastTrack->mHapticIntensity = os::HapticScale::NONE;
+        fastTrack->mHapticMaxAmplitude = NAN;
         fastTrack->mGeneration++;
         state->mFastTracksGen++;
         state->mTrackMask = 1;
@@ -4861,6 +4980,9 @@
         // before effects processing or output.
         if (mMixerBufferValid) {
             memset(mMixerBuffer, 0, mMixerBufferSize);
+            if (mType == SPATIALIZER) {
+                memset(mSinkBuffer, 0, mSinkBufferSize);
+            }
         } else {
             memset(mSinkBuffer, 0, mSinkBufferSize);
         }
@@ -5103,6 +5225,7 @@
                     fastTrack->mFormat = track->mFormat;
                     fastTrack->mHapticPlaybackEnabled = track->getHapticPlaybackEnabled();
                     fastTrack->mHapticIntensity = track->getHapticIntensity();
+                    fastTrack->mHapticMaxAmplitude = track->getHapticMaxAmplitude();
                     fastTrack->mGeneration++;
                     state->mTrackMask |= 1 << j;
                     didModify = true;
@@ -5352,11 +5475,21 @@
                 trackId,
                 AudioMixer::TRACK,
                 AudioMixer::CHANNEL_MASK, (void *)(uintptr_t)track->channelMask());
-            mAudioMixer->setParameter(
-                trackId,
-                AudioMixer::TRACK,
-                AudioMixer::MIXER_CHANNEL_MASK,
-                (void *)(uintptr_t)(mChannelMask | mHapticChannelMask));
+
+            if (mType == SPATIALIZER && !track->canBeSpatialized()) {
+                mAudioMixer->setParameter(
+                    trackId,
+                    AudioMixer::TRACK,
+                    AudioMixer::MIXER_CHANNEL_MASK,
+                    (void *)(uintptr_t)(mChannelMask | mHapticChannelMask));
+            } else {
+                mAudioMixer->setParameter(
+                    trackId,
+                    AudioMixer::TRACK,
+                    AudioMixer::MIXER_CHANNEL_MASK,
+                    (void *)(uintptr_t)(mMixerChannelMask | mHapticChannelMask));
+            }
+
             // limit track sample rate to 2 x output sample rate, which changes at re-configuration
             uint32_t maxSampleRate = mSampleRate * AUDIO_RESAMPLER_DOWN_RATIO_MAX;
             uint32_t reqSampleRate = proxy->getSampleRate();
@@ -5393,16 +5526,27 @@
             if (mMixerBufferEnabled
                     && (track->mainBuffer() == mSinkBuffer
                             || track->mainBuffer() == mMixerBuffer)) {
-                mAudioMixer->setParameter(
-                        trackId,
-                        AudioMixer::TRACK,
-                        AudioMixer::MIXER_FORMAT, (void *)mMixerBufferFormat);
-                mAudioMixer->setParameter(
-                        trackId,
-                        AudioMixer::TRACK,
-                        AudioMixer::MAIN_BUFFER, (void *)mMixerBuffer);
-                // TODO: override track->mainBuffer()?
-                mMixerBufferValid = true;
+                if (mType == SPATIALIZER && !track->canBeSpatialized()) {
+                    mAudioMixer->setParameter(
+                            trackId,
+                            AudioMixer::TRACK,
+                            AudioMixer::MIXER_FORMAT, (void *)mFormat);
+                    mAudioMixer->setParameter(
+                            trackId,
+                            AudioMixer::TRACK,
+                            AudioMixer::MAIN_BUFFER, (void *)mSinkBuffer);
+                } else {
+                    mAudioMixer->setParameter(
+                            trackId,
+                            AudioMixer::TRACK,
+                            AudioMixer::MIXER_FORMAT, (void *)mMixerBufferFormat);
+                    mAudioMixer->setParameter(
+                            trackId,
+                            AudioMixer::TRACK,
+                            AudioMixer::MAIN_BUFFER, (void *)mMixerBuffer);
+                    // TODO: override track->mainBuffer()?
+                    mMixerBufferValid = true;
+                }
             } else {
                 mAudioMixer->setParameter(
                         trackId,
@@ -5425,6 +5569,10 @@
                 trackId,
                 AudioMixer::TRACK,
                 AudioMixer::HAPTIC_INTENSITY, (void *)(uintptr_t)track->getHapticIntensity());
+            mAudioMixer->setParameter(
+                trackId,
+                AudioMixer::TRACK,
+                AudioMixer::HAPTIC_MAX_AMPLITUDE, (void *)(&(track->mHapticMaxAmplitude)));
 
             // reset retry count
             track->mRetryCount = kMaxTrackRetries;
@@ -5575,7 +5723,8 @@
     // remove all the tracks that need to be...
     removeTracks_l(*tracksToRemove);
 
-    if (getEffectChain_l(AUDIO_SESSION_OUTPUT_MIX) != 0) {
+    if (getEffectChain_l(AUDIO_SESSION_OUTPUT_MIX) != 0 ||
+            getEffectChain_l(AUDIO_SESSION_OUTPUT_STAGE) != 0) {
         mEffectBufferValid = true;
     }
 
@@ -5587,8 +5736,10 @@
     // sink or mix buffer must be cleared if all tracks are connected to an
     // effect chain as in this case the mixer will not write to the sink or mix buffer
     // and track effects will accumulate into it
-    if ((mBytesRemaining == 0) && ((mixedTracks != 0 && mixedTracks == tracksWithEffect) ||
-            (mixedTracks == 0 && fastTracks > 0))) {
+    // always clear sink buffer for spatializer output as the output of the spatializer
+    // effect will be accumulated into it
+    if ((mBytesRemaining == 0) && (((mixedTracks != 0 && mixedTracks == tracksWithEffect) ||
+            (mixedTracks == 0 && fastTracks > 0)) || (mType == SPATIALIZER))) {
         // FIXME as a performance optimization, should remember previous zero status
         if (mMixerBufferValid) {
             memset(mMixerBuffer, 0, mMixerBufferSize);
@@ -6971,6 +7122,69 @@
     MixerThread::cacheParameters_l();
 }
 
+// ----------------------------------------------------------------------------
+
+AudioFlinger::SpatializerThread::SpatializerThread(const sp<AudioFlinger>& audioFlinger,
+                                                             AudioStreamOut* output,
+                                                             audio_io_handle_t id,
+                                                             bool systemReady,
+                                                             audio_config_base_t *mixerConfig)
+    : MixerThread(audioFlinger, output, id, systemReady, SPATIALIZER, mixerConfig)
+{
+}
+
+void AudioFlinger::SpatializerThread::checkOutputStageEffects()
+{
+    bool hasVirtualizer = false;
+    bool hasDownMixer = false;
+    sp<EffectHandle> finalDownMixer;
+    {
+        Mutex::Autolock _l(mLock);
+        sp<EffectChain> chain = getEffectChain_l(AUDIO_SESSION_OUTPUT_STAGE);
+        if (chain != 0) {
+            hasVirtualizer = chain->getEffectFromType_l(FX_IID_SPATIALIZER) != nullptr;
+            hasDownMixer = chain->getEffectFromType_l(EFFECT_UIID_DOWNMIX) != nullptr;
+        }
+
+        finalDownMixer = mFinalDownMixer;
+        mFinalDownMixer.clear();
+    }
+
+    if (hasVirtualizer) {
+        if (finalDownMixer != nullptr) {
+            int32_t ret;
+            finalDownMixer->disable(&ret);
+        }
+        finalDownMixer.clear();
+    } else if (!hasDownMixer) {
+        std::vector<effect_descriptor_t> descriptors;
+        status_t status = mAudioFlinger->mEffectsFactoryHal->getDescriptors(
+                                                        EFFECT_UIID_DOWNMIX, &descriptors);
+        if (status != NO_ERROR) {
+            return;
+        }
+        ALOG_ASSERT(!descriptors.empty(),
+                "%s getDescriptors() returned no error but empty list", __func__);
+
+        finalDownMixer = createEffect_l(nullptr /*client*/, nullptr /*effectClient*/,
+                0 /*priority*/, AUDIO_SESSION_OUTPUT_STAGE, &descriptors[0], nullptr /*enabled*/,
+                &status, false /*pinned*/, false /*probe*/, false /*notifyFramesProcessed*/);
+
+        if (finalDownMixer == nullptr || (status != NO_ERROR && status != ALREADY_EXISTS)) {
+            ALOGW("%s error creating downmixer %d", __func__, status);
+            finalDownMixer.clear();
+        } else {
+            int32_t ret;
+            finalDownMixer->enable(&ret);
+        }
+    }
+
+    {
+        Mutex::Autolock _l(mLock);
+        mFinalDownMixer = finalDownMixer;
+    }
+}
+
 
 // ----------------------------------------------------------------------------
 //      Record
@@ -9278,7 +9492,7 @@
     mActiveTracks.add(track);
     sp<EffectChain> chain = getEffectChain_l(mSessionId);
     if (chain != 0) {
-        chain->setStrategy(AudioSystem::getStrategyForStream(streamType()));
+        chain->setStrategy(getStrategyForStream(streamType()));
         chain->incTrackCnt();
         chain->incActiveTrackCnt();
     }
diff --git a/services/audioflinger/Threads.h b/services/audioflinger/Threads.h
index 16082a9..0e86391 100644
--- a/services/audioflinger/Threads.h
+++ b/services/audioflinger/Threads.h
@@ -32,6 +32,7 @@
         OFFLOAD,            // Thread class is OffloadThread
         MMAP_PLAYBACK,      // Thread class for MMAP playback stream
         MMAP_CAPTURE,       // Thread class for MMAP capture stream
+        SPATIALIZER,  //
         // If you add any values here, also update ThreadBase::threadTypeToString()
     };
 
@@ -53,7 +54,8 @@
         CFG_EVENT_CREATE_AUDIO_PATCH,
         CFG_EVENT_RELEASE_AUDIO_PATCH,
         CFG_EVENT_UPDATE_OUT_DEVICE,
-        CFG_EVENT_RESIZE_BUFFER
+        CFG_EVENT_RESIZE_BUFFER,
+        CFG_EVENT_CHECK_OUTPUT_STAGE_EFFECTS
     };
 
     class ConfigEventData: public RefBase {
@@ -87,7 +89,13 @@
     public:
         virtual ~ConfigEvent() {}
 
-        void dump(char *buffer, size_t size) { mData->dump(buffer, size); }
+        void dump(char *buffer, size_t size) {
+            snprintf(buffer, size, "Event type: %d\n", mType);
+            if (mData != nullptr) {
+                snprintf(buffer, size, "Data:\n");
+                mData->dump(buffer, size);
+            }
+        }
 
         const int mType; // event type e.g. CFG_EVENT_IO
         Mutex mLock;     // mutex associated with mCond
@@ -110,7 +118,7 @@
             mEvent(event), mPid(pid), mPortId(portId) {}
 
         virtual  void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "IO event: event %d\n", mEvent);
+            snprintf(buffer, size, "- IO event: event %d\n", mEvent);
         }
 
         const audio_io_config_event mEvent;
@@ -133,7 +141,7 @@
             mPid(pid), mTid(tid), mPrio(prio), mForApp(forApp) {}
 
         virtual  void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "Prio event: pid %d, tid %d, prio %d, for app? %d\n",
+            snprintf(buffer, size, "- Prio event: pid %d, tid %d, prio %d, for app? %d\n",
                     mPid, mTid, mPrio, mForApp);
         }
 
@@ -158,7 +166,7 @@
             mKeyValuePairs(keyValuePairs) {}
 
         virtual  void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "KeyValue: %s\n", mKeyValuePairs.string());
+            snprintf(buffer, size, "- KeyValue: %s\n", mKeyValuePairs.string());
         }
 
         const String8 mKeyValuePairs;
@@ -181,7 +189,7 @@
             mPatch(patch), mHandle(handle) {}
 
         virtual  void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "Patch handle: %u\n", mHandle);
+            snprintf(buffer, size, "- Patch handle: %u\n", mHandle);
         }
 
         const struct audio_patch mPatch;
@@ -205,7 +213,7 @@
             mHandle(handle) {}
 
         virtual  void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "Patch handle: %u\n", mHandle);
+            snprintf(buffer, size, "- Patch handle: %u\n", mHandle);
         }
 
         audio_patch_handle_t mHandle;
@@ -227,7 +235,7 @@
             mOutDevices(outDevices) {}
 
         virtual void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "Devices: %s", android::toString(mOutDevices).c_str());
+            snprintf(buffer, size, "- Devices: %s", android::toString(mOutDevices).c_str());
         }
 
         DeviceDescriptorBaseVector mOutDevices;
@@ -249,7 +257,7 @@
             mMaxSharedAudioHistoryMs(maxSharedAudioHistoryMs) {}
 
         virtual void dump(char *buffer, size_t size) {
-            snprintf(buffer, size, "mMaxSharedAudioHistoryMs: %d", mMaxSharedAudioHistoryMs);
+            snprintf(buffer, size, "- mMaxSharedAudioHistoryMs: %d", mMaxSharedAudioHistoryMs);
         }
 
         int32_t mMaxSharedAudioHistoryMs;
@@ -265,6 +273,16 @@
         virtual ~ResizeBufferConfigEvent() {}
     };
 
+    class CheckOutputStageEffectsEvent : public ConfigEvent {
+    public:
+        CheckOutputStageEffectsEvent() :
+            ConfigEvent(CFG_EVENT_CHECK_OUTPUT_STAGE_EFFECTS) {
+        }
+
+        virtual ~CheckOutputStageEffectsEvent() {}
+    };
+
+
     class PMDeathRecipient : public IBinder::DeathRecipient {
     public:
         explicit    PMDeathRecipient(const wp<ThreadBase>& thread) : mThread(thread) {}
@@ -290,8 +308,11 @@
                 // dynamic externally-visible
                 uint32_t    sampleRate() const { return mSampleRate; }
                 audio_channel_mask_t channelMask() const { return mChannelMask; }
+    virtual     audio_channel_mask_t mixerChannelMask() const { return mChannelMask; }
+
                 audio_format_t format() const { return mHALFormat; }
                 uint32_t channelCount() const { return mChannelCount; }
+
                 // Called by AudioFlinger::frameCount(audio_io_handle_t output) and effects,
                 // and returns the [normal mix] buffer's frame count.
     virtual     size_t      frameCount() const = 0;
@@ -330,7 +351,11 @@
                 status_t    sendUpdateOutDeviceConfigEvent(
                                     const DeviceDescriptorBaseVector& outDevices);
                 void        sendResizeBufferConfigEvent_l(int32_t maxSharedAudioHistoryMs);
+                void        sendCheckOutputStageEffectsEvent();
+                void        sendCheckOutputStageEffectsEvent_l();
+
                 void        processConfigEvents_l();
+    virtual     void        setCheckOutputStageEffects() {}
     virtual     void        cacheParameters_l() = 0;
     virtual     status_t    createAudioPatch_l(const struct audio_patch *patch,
                                                audio_patch_handle_t *handle) = 0;
@@ -383,7 +408,8 @@
                                     int *enabled,
                                     status_t *status /*non-NULL*/,
                                     bool pinned,
-                                    bool probe);
+                                    bool probe,
+                                    bool notifyFramesProcessed);
 
                 // return values for hasAudioSession (bit field)
                 enum effect_state {
@@ -574,6 +600,8 @@
                                 return INVALID_OPERATION;
                             }
 
+                product_strategy_t getStrategyForStream(audio_stream_type_t stream) const;
+
     virtual     void        dumpInternals_l(int fd __unused, const Vector<String16>& args __unused)
                             { }
     virtual     void        dumpTracks_l(int fd __unused, const Vector<String16>& args __unused) { }
@@ -824,7 +852,8 @@
     static const nsecs_t kMaxNextBufferDelayNs = 100000000;
 
     PlaybackThread(const sp<AudioFlinger>& audioFlinger, AudioStreamOut* output,
-                   audio_io_handle_t id, type_t type, bool systemReady);
+                   audio_io_handle_t id, type_t type, bool systemReady,
+                   audio_config_base_t *mixerConfig = nullptr);
     virtual             ~PlaybackThread();
 
     // Thread virtuals
@@ -881,6 +910,8 @@
                                 mActiveTracks.updatePowerState(this, true /* force */);
                             }
 
+    virtual     void        checkOutputStageEffects() {}
+
                 void        dumpInternals_l(int fd, const Vector<String16>& args) override;
                 void        dumpTracks_l(int fd, const Vector<String16>& args) override;
 
@@ -973,6 +1004,10 @@
 
     virtual     size_t      frameCount() const { return mNormalFrameCount; }
 
+                audio_channel_mask_t mixerChannelMask() const override {
+                    return mMixerChannelMask;
+                }
+
                 status_t    getTimestamp_l(AudioTimestamp& timestamp);
 
                 void        addPatchTrack(const sp<PatchTrack>& track);
@@ -1015,6 +1050,9 @@
 
                 PlaybackThread::Track* getTrackById_l(audio_port_handle_t trackId);
 
+                bool hasMixer() const {
+                    return mType == MIXER || mType == DUPLICATING || mType == SPATIALIZER;
+                }
 protected:
     // updated by readOutputParameters_l()
     size_t                          mNormalFrameCount;  // normal mixer and effects
@@ -1084,6 +1122,11 @@
     // for any processing (including output processing).
     bool                            mEffectBufferValid;
 
+    // Frame size aligned buffer used to convert mEffectBuffer samples to mSinkBuffer format prior
+    // to accumulate into mSinkBuffer on SPATIALIZER threads
+    void*                           mEffectToSinkBuffer = nullptr;
+
+
     // suspend count, > 0 means suspended.  While suspended, the thread continues to pull from
     // tracks and mix, but doesn't write to HAL.  A2DP and SCO HAL implementations can't handle
     // concurrent use of both of them, so Audio Policy Service suspends one of the threads to
@@ -1101,6 +1144,9 @@
     // haptic playback.
     audio_channel_mask_t            mHapticChannelMask = AUDIO_CHANNEL_NONE;
     uint32_t                        mHapticChannelCount = 0;
+
+    audio_channel_mask_t            mMixerChannelMask = AUDIO_CHANNEL_NONE;
+
 private:
     // mMasterMute is in both PlaybackThread and in AudioFlinger.  When a
     // PlaybackThread needs to find out if master-muted, it checks it's local
@@ -1134,6 +1180,9 @@
 
     // Cache various calculated values, at threadLoop() entry and after a parameter change
     virtual     void        cacheParameters_l();
+                void        setCheckOutputStageEffects() override {
+                                mCheckOutputStageEffects.store(true);
+                            }
 
     virtual     uint32_t    correctLatency_l(uint32_t latency) const;
 
@@ -1314,6 +1363,8 @@
                 // audio patch used by the downstream software patch.
                 // Only used if ThreadBase::mIsMsdDevice is true.
                 struct audio_patch mDownStreamPatch;
+
+                std::atomic_bool mCheckOutputStageEffects{};
 };
 
 class MixerThread : public PlaybackThread {
@@ -1322,7 +1373,8 @@
                 AudioStreamOut* output,
                 audio_io_handle_t id,
                 bool systemReady,
-                type_t type = MIXER);
+                type_t type = MIXER,
+                audio_config_base_t *mixerConfig = nullptr);
     virtual             ~MixerThread();
 
     // Thread virtuals
@@ -1611,6 +1663,24 @@
     }
 };
 
+class SpatializerThread : public MixerThread {
+public:
+    SpatializerThread(const sp<AudioFlinger>& audioFlinger,
+                           AudioStreamOut* output,
+                           audio_io_handle_t id,
+                           bool systemReady,
+                           audio_config_base_t *mixerConfig);
+            ~SpatializerThread() override {}
+
+            bool hasFastMixer() const override { return false; }
+
+protected:
+            void checkOutputStageEffects() override;
+
+private:
+            sp<EffectHandle> mFinalDownMixer;
+};
+
 // record thread
 class RecordThread : public ThreadBase
 {
diff --git a/services/audiopolicy/AudioPolicyInterface.h b/services/audiopolicy/AudioPolicyInterface.h
index 2e49e71..4078278 100644
--- a/services/audiopolicy/AudioPolicyInterface.h
+++ b/services/audiopolicy/AudioPolicyInterface.h
@@ -333,6 +333,50 @@
     virtual status_t getDevicesForRoleAndCapturePreset(audio_source_t audioSource,
                                                        device_role_t role,
                                                        AudioDeviceTypeAddrVector &devices) = 0;
+
+    /**
+     * Queries if some kind of spatialization will be performed if the audio playback context
+     * described by the provided arguments is present.
+     * The context is made of:
+     * - The audio attributes describing the playback use case.
+     * - The audio configuration describing the audio format, channels, sampling rate ...
+     * - The devices describing the sink audio device selected for playback.
+     * All arguments are optional and only the specified arguments are used to match against
+     * supported criteria. For instance, supplying no argument will tell if spatialization is
+     * supported or not in general.
+     * @param attr audio attributes describing the playback use case
+     * @param config audio configuration describing the audio format, channels, sampling rate...
+     * @param devices the sink audio device selected for playback
+     * @return true if spatialization is enabled for this context,
+     *        false otherwise
+     */
+     virtual bool canBeSpatialized(const audio_attributes_t *attr,
+                                  const audio_config_t *config,
+                                  const AudioDeviceTypeAddrVector &devices) const = 0;
+
+    /**
+     * Opens a specialized spatializer output if supported by the platform.
+     * If several spatializer output profiles exist, the one supporting the sink device
+     * corresponding to the provided audio attributes will be selected.
+     * Only one spatializer output stream can be opened at a time and an error is returned
+     * if one already exists.
+     * @param config audio format, channel mask and sampling rate to be used as the mixer
+     *        configuration for the spatializer mixer created.
+     * @param attr audio attributes describing the playback use case that will drive the
+     *        sink device selection
+     * @param output the IO handle of the output opened
+     * @return NO_ERROR if an output was opened, INVALID_OPERATION or BAD_VALUE otherwise
+     */
+    virtual status_t getSpatializerOutput(const audio_config_base_t *config,
+                                            const audio_attributes_t *attr,
+                                            audio_io_handle_t *output) = 0;
+
+    /**
+     * Closes a previously opened specialized spatializer output.
+     * @param output the IO handle of the output to close.
+     * @return NO_ERROR if an output was closed, INVALID_OPERATION or BAD_VALUE otherwise
+     */
+    virtual status_t releaseSpatializerOutput(audio_io_handle_t output) = 0;
 };
 
 
@@ -359,7 +403,8 @@
     // The audio policy manager can check if the proposed parameters are suitable or not and act accordingly.
     virtual status_t openOutput(audio_module_handle_t module,
                                 audio_io_handle_t *output,
-                                audio_config_t *config,
+                                audio_config_t *halConfig,
+                                audio_config_base_t *mixerConfig,
                                 const sp<DeviceDescriptorBase>& device,
                                 uint32_t *latencyMs,
                                 audio_output_flags_t flags) = 0;
diff --git a/services/audiopolicy/common/managerdefinitions/include/AudioOutputDescriptor.h b/services/audiopolicy/common/managerdefinitions/include/AudioOutputDescriptor.h
index 1f9b535..7c7f02d 100644
--- a/services/audiopolicy/common/managerdefinitions/include/AudioOutputDescriptor.h
+++ b/services/audiopolicy/common/managerdefinitions/include/AudioOutputDescriptor.h
@@ -362,7 +362,8 @@
                            const struct audio_port_config *srcConfig = NULL) const;
     virtual void toAudioPort(struct audio_port_v7 *port) const;
 
-        status_t open(const audio_config_t *config,
+        status_t open(const audio_config_t *halConfig,
+                      const audio_config_base_t *mixerConfig,
                       const DeviceVector &devices,
                       audio_stream_type_t stream,
                       audio_output_flags_t flags,
@@ -423,6 +424,7 @@
     uint32_t mDirectOpenCount; // number of clients using this output (direct outputs only)
     audio_session_t mDirectClientSession; // session id of the direct output client
     bool mPendingReopenToQueryProfiles = false;
+    audio_channel_mask_t mMixerChannelMask = AUDIO_CHANNEL_NONE;
 };
 
 // Audio output driven by an input device directly.
diff --git a/services/audiopolicy/common/managerdefinitions/include/AudioPolicyConfig.h b/services/audiopolicy/common/managerdefinitions/include/AudioPolicyConfig.h
index cf1f64c..a8fd856 100644
--- a/services/audiopolicy/common/managerdefinitions/include/AudioPolicyConfig.h
+++ b/services/audiopolicy/common/managerdefinitions/include/AudioPolicyConfig.h
@@ -202,6 +202,20 @@
             {AUDIO_FORMAT_AC4, {}}};
     }
 
+    //TODO: b/193496180 use spatializer flag at audio HAL when available
+    // until then, use DEEP_BUFFER+FAST flag combo to indicate the spatializer output profile
+    void convertSpatializerFlag()
+    {
+        for (const auto& hwModule : mHwModules) {
+            for (const auto& curProfile : hwModule->getOutputProfiles()) {
+                if (curProfile->getFlags()
+                        == (AUDIO_OUTPUT_FLAG_FAST | AUDIO_OUTPUT_FLAG_DEEP_BUFFER)) {
+                    curProfile->setFlags(AUDIO_OUTPUT_FLAG_SPATIALIZER);
+                }
+            }
+        }
+    }
+
 private:
     static const constexpr char* const kDefaultEngineLibraryNameSuffix = "default";
 
diff --git a/services/audiopolicy/common/managerdefinitions/include/DeviceDescriptor.h b/services/audiopolicy/common/managerdefinitions/include/DeviceDescriptor.h
index 20b4044..58d05c6 100644
--- a/services/audiopolicy/common/managerdefinitions/include/DeviceDescriptor.h
+++ b/services/audiopolicy/common/managerdefinitions/include/DeviceDescriptor.h
@@ -168,6 +168,10 @@
     DeviceVector getDevicesFromDeviceTypeAddrVec(
             const AudioDeviceTypeAddrVector& deviceTypeAddrVector) const;
 
+    // Return the device vector that contains device descriptor whose AudioDeviceTypeAddr appears
+    // in the given AudioDeviceTypeAddrVector
+    AudioDeviceTypeAddrVector toTypeAddrVector() const;
+
     // If there are devices with the given type and the devices to add is not empty,
     // remove all the devices with the given type and add all the devices to add.
     void replaceDevicesByType(audio_devices_t typeToRemove, const DeviceVector &devicesToAdd);
diff --git a/services/audiopolicy/common/managerdefinitions/src/AudioOutputDescriptor.cpp b/services/audiopolicy/common/managerdefinitions/src/AudioOutputDescriptor.cpp
index 6b08f7c..f3d2326 100644
--- a/services/audiopolicy/common/managerdefinitions/src/AudioOutputDescriptor.cpp
+++ b/services/audiopolicy/common/managerdefinitions/src/AudioOutputDescriptor.cpp
@@ -491,7 +491,8 @@
     return true;
 }
 
-status_t SwAudioOutputDescriptor::open(const audio_config_t *config,
+status_t SwAudioOutputDescriptor::open(const audio_config_t *halConfig,
+                                       const audio_config_base_t *mixerConfig,
                                        const DeviceVector &devices,
                                        audio_stream_type_t stream,
                                        audio_output_flags_t flags,
@@ -504,45 +505,62 @@
                         "with the requested devices, all device types: %s",
                         __func__, dumpDeviceTypes(devices.types()).c_str());
 
-    audio_config_t lConfig;
-    if (config == nullptr) {
-        lConfig = AUDIO_CONFIG_INITIALIZER;
-        lConfig.sample_rate = mSamplingRate;
-        lConfig.channel_mask = mChannelMask;
-        lConfig.format = mFormat;
+    audio_config_t lHalConfig;
+    if (halConfig == nullptr) {
+        lHalConfig = AUDIO_CONFIG_INITIALIZER;
+        lHalConfig.sample_rate = mSamplingRate;
+        lHalConfig.channel_mask = mChannelMask;
+        lHalConfig.format = mFormat;
     } else {
-        lConfig = *config;
+        lHalConfig = *halConfig;
     }
 
     // if the selected profile is offloaded and no offload info was specified,
     // create a default one
     if ((mProfile->getFlags() & AUDIO_OUTPUT_FLAG_COMPRESS_OFFLOAD) &&
-            lConfig.offload_info.format == AUDIO_FORMAT_DEFAULT) {
+            lHalConfig.offload_info.format == AUDIO_FORMAT_DEFAULT) {
         flags = (audio_output_flags_t)(flags | AUDIO_OUTPUT_FLAG_COMPRESS_OFFLOAD);
-        lConfig.offload_info = AUDIO_INFO_INITIALIZER;
-        lConfig.offload_info.sample_rate = lConfig.sample_rate;
-        lConfig.offload_info.channel_mask = lConfig.channel_mask;
-        lConfig.offload_info.format = lConfig.format;
-        lConfig.offload_info.stream_type = stream;
-        lConfig.offload_info.duration_us = -1;
-        lConfig.offload_info.has_video = true; // conservative
-        lConfig.offload_info.is_streaming = true; // likely
-        lConfig.offload_info.encapsulation_mode = lConfig.offload_info.encapsulation_mode;
-        lConfig.offload_info.content_id = lConfig.offload_info.content_id;
-        lConfig.offload_info.sync_id = lConfig.offload_info.sync_id;
+        lHalConfig.offload_info = AUDIO_INFO_INITIALIZER;
+        lHalConfig.offload_info.sample_rate = lHalConfig.sample_rate;
+        lHalConfig.offload_info.channel_mask = lHalConfig.channel_mask;
+        lHalConfig.offload_info.format = lHalConfig.format;
+        lHalConfig.offload_info.stream_type = stream;
+        lHalConfig.offload_info.duration_us = -1;
+        lHalConfig.offload_info.has_video = true; // conservative
+        lHalConfig.offload_info.is_streaming = true; // likely
+        lHalConfig.offload_info.encapsulation_mode = lHalConfig.offload_info.encapsulation_mode;
+        lHalConfig.offload_info.content_id = lHalConfig.offload_info.content_id;
+        lHalConfig.offload_info.sync_id = lHalConfig.offload_info.sync_id;
+    }
+
+    audio_config_base_t lMixerConfig;
+    if (mixerConfig == nullptr) {
+        lMixerConfig = AUDIO_CONFIG_BASE_INITIALIZER;
+        lMixerConfig.sample_rate = lHalConfig.sample_rate;
+        lMixerConfig.channel_mask = lHalConfig.channel_mask;
+        lMixerConfig.format = lHalConfig.format;
+    } else {
+        lMixerConfig = *mixerConfig;
     }
 
     mFlags = (audio_output_flags_t)(mFlags | flags);
 
+    //TODO: b/193496180 use spatializer flag at audio HAL when available
+    audio_output_flags_t halFlags = mFlags;
+    if ((mFlags & AUDIO_OUTPUT_FLAG_SPATIALIZER) != 0) {
+        halFlags = (audio_output_flags_t)(AUDIO_OUTPUT_FLAG_FAST | AUDIO_OUTPUT_FLAG_DEEP_BUFFER);
+    }
+
     ALOGV("opening output for device %s profile %p name %s",
           mDevices.toString().c_str(), mProfile.get(), mProfile->getName().c_str());
 
     status_t status = mClientInterface->openOutput(mProfile->getModuleHandle(),
                                                    output,
-                                                   &lConfig,
+                                                   &lHalConfig,
+                                                   &lMixerConfig,
                                                    device,
                                                    &mLatency,
-                                                   mFlags);
+                                                   halFlags);
 
     if (status == NO_ERROR) {
         LOG_ALWAYS_FATAL_IF(*output == AUDIO_IO_HANDLE_NONE,
@@ -550,9 +568,10 @@
                             "selected device %s for opening",
                             __FUNCTION__, *output, devices.toString().c_str(),
                             device->toString().c_str());
-        mSamplingRate = lConfig.sample_rate;
-        mChannelMask = lConfig.channel_mask;
-        mFormat = lConfig.format;
+        mSamplingRate = lHalConfig.sample_rate;
+        mChannelMask = lHalConfig.channel_mask;
+        mFormat = lHalConfig.format;
+        mMixerChannelMask = lMixerConfig.channel_mask;
         mId = PolicyAudioPort::getNextUniqueId();
         mIoHandle = *output;
         mProfile->curOpenCount++;
diff --git a/services/audiopolicy/common/managerdefinitions/src/DeviceDescriptor.cpp b/services/audiopolicy/common/managerdefinitions/src/DeviceDescriptor.cpp
index a92d31e..1722032 100644
--- a/services/audiopolicy/common/managerdefinitions/src/DeviceDescriptor.cpp
+++ b/services/audiopolicy/common/managerdefinitions/src/DeviceDescriptor.cpp
@@ -451,6 +451,14 @@
     return devices;
 }
 
+AudioDeviceTypeAddrVector DeviceVector::toTypeAddrVector() const {
+    AudioDeviceTypeAddrVector result;
+    for (const auto& device : *this) {
+        result.push_back(AudioDeviceTypeAddr(device->type(), device->address()));
+    }
+    return result;
+}
+
 void DeviceVector::replaceDevicesByType(
         audio_devices_t typeToRemove, const DeviceVector &devicesToAdd) {
     DeviceVector devicesToRemove = getDevicesFromType(typeToRemove);
diff --git a/services/audiopolicy/fuzzer/audiopolicy_fuzzer.cpp b/services/audiopolicy/fuzzer/audiopolicy_fuzzer.cpp
index 7000cd9..8584702 100644
--- a/services/audiopolicy/fuzzer/audiopolicy_fuzzer.cpp
+++ b/services/audiopolicy/fuzzer/audiopolicy_fuzzer.cpp
@@ -163,7 +163,9 @@
     AUDIO_FLAG_BYPASS_MUTE,    AUDIO_FLAG_LOW_LATENCY,
     AUDIO_FLAG_DEEP_BUFFER,    AUDIO_FLAG_NO_MEDIA_PROJECTION,
     AUDIO_FLAG_MUTE_HAPTIC,    AUDIO_FLAG_NO_SYSTEM_CAPTURE,
-    AUDIO_FLAG_CAPTURE_PRIVATE};
+    AUDIO_FLAG_CAPTURE_PRIVATE, AUDIO_FLAG_CONTENT_SPATIALIZED,
+    AUDIO_FLAG_NEVER_SPATIALIZE,
+    };
 
 std::vector<audio_policy_dev_state_t> kAudioPolicyDeviceStates = {
     AUDIO_POLICY_DEVICE_STATE_UNAVAILABLE,
diff --git a/services/audiopolicy/managerdefault/AudioPolicyManager.cpp b/services/audiopolicy/managerdefault/AudioPolicyManager.cpp
index cb9d700..e334532 100644
--- a/services/audiopolicy/managerdefault/AudioPolicyManager.cpp
+++ b/services/audiopolicy/managerdefault/AudioPolicyManager.cpp
@@ -246,8 +246,8 @@
                     sp<SwAudioOutputDescriptor> desc = mOutputs.valueFor(output);
                     // close unused outputs after device disconnection or direct outputs that have
                     // been opened by checkOutputsForDevice() to query dynamic parameters
-                    if ((state == AUDIO_POLICY_DEVICE_STATE_UNAVAILABLE) ||
-                            (((desc->mFlags & AUDIO_OUTPUT_FLAG_DIRECT) != 0) &&
+                    if ((state == AUDIO_POLICY_DEVICE_STATE_UNAVAILABLE)
+                            || (((desc->mFlags & AUDIO_OUTPUT_FLAG_DIRECT) != 0) &&
                                 (desc->mDirectOpenCount == 0))) {
                         clearAudioSourcesForOutput(output);
                         closeOutput(output);
@@ -925,6 +925,32 @@
     return profile;
 }
 
+sp<IOProfile> AudioPolicyManager::getSpatializerOutputProfile(
+        const audio_config_t *config __unused, const AudioDeviceTypeAddrVector &devices) const
+{
+    for (const auto& hwModule : mHwModules) {
+        for (const auto& curProfile : hwModule->getOutputProfiles()) {
+            if (curProfile->getFlags() != AUDIO_OUTPUT_FLAG_SPATIALIZER) {
+                continue;
+            }
+            // reject profiles not corresponding to a device currently available
+            DeviceVector supportedDevices = curProfile->getSupportedDevices();
+            if (!mAvailableOutputDevices.containsAtLeastOne(supportedDevices)) {
+                continue;
+            }
+            if (!devices.empty()) {
+                if (supportedDevices.getDevicesFromDeviceTypeAddrVec(devices).size()
+                        != devices.size()) {
+                    continue;
+                }
+            }
+            ALOGV("%s found profile %s", __func__, curProfile->getName().c_str());
+            return curProfile;
+        }
+    }
+    return nullptr;
+}
+
 audio_io_handle_t AudioPolicyManager::getOutput(audio_stream_type_t stream)
 {
     DeviceVector devices = mEngine->getOutputDevicesForStream(stream, false /*fromCache*/);
@@ -1094,7 +1120,7 @@
 
     *output = AUDIO_IO_HANDLE_NONE;
     if (!msdDevices.isEmpty()) {
-        *output = getOutputForDevices(msdDevices, session, *stream, config, flags);
+        *output = getOutputForDevices(msdDevices, session, resultAttr, config, flags);
         if (*output != AUDIO_IO_HANDLE_NONE && setMsdOutputPatches(&outputDevices) == NO_ERROR) {
             ALOGV("%s() Using MSD devices %s instead of devices %s",
                   __func__, msdDevices.toString().c_str(), outputDevices.toString().c_str());
@@ -1103,7 +1129,7 @@
         }
     }
     if (*output == AUDIO_IO_HANDLE_NONE) {
-        *output = getOutputForDevices(outputDevices, session, *stream, config,
+        *output = getOutputForDevices(outputDevices, session, resultAttr, config,
                 flags, resultAttr->flags & AUDIO_FLAG_MUTE_HAPTIC);
     }
     if (*output == AUDIO_IO_HANDLE_NONE) {
@@ -1265,7 +1291,8 @@
     // all MSD patches to prioritize this request over any active output on MSD.
     releaseMsdOutputPatches(devices);
 
-    status_t status = outputDesc->open(config, devices, stream, flags, output);
+    status_t status =
+            outputDesc->open(config, nullptr /* mixerConfig */, devices, stream, flags, output);
 
     // only accept an output with the requested parameters
     if (status != NO_ERROR ||
@@ -1300,7 +1327,7 @@
 audio_io_handle_t AudioPolicyManager::getOutputForDevices(
         const DeviceVector &devices,
         audio_session_t session,
-        audio_stream_type_t stream,
+        const audio_attributes_t *attr,
         const audio_config_t *config,
         audio_output_flags_t *flags,
         bool forceMutingHaptic)
@@ -1322,6 +1349,9 @@
     if ((*flags & AUDIO_OUTPUT_FLAG_HW_AV_SYNC) != 0) {
         *flags = (audio_output_flags_t)(*flags | AUDIO_OUTPUT_FLAG_DIRECT);
     }
+
+    audio_stream_type_t stream = mEngine->getStreamTypeForAttributes(*attr);
+
     // only allow deep buffering for music stream type
     if (stream != AUDIO_STREAM_MUSIC) {
         *flags = (audio_output_flags_t)(*flags &~AUDIO_OUTPUT_FLAG_DEEP_BUFFER);
@@ -1341,6 +1371,11 @@
         ALOGV("Set VoIP and Direct output flags for PCM format");
     }
 
+    if (mSpatializerOutput != nullptr
+            && canBeSpatialized(attr, config, devices.toTypeAddrVector())) {
+        return mSpatializerOutput->mIoHandle;
+    }
+
     audio_config_t directConfig = *config;
     directConfig.channel_mask = channelMask;
     status_t status = openDirectOutput(stream, session, &directConfig, *flags, devices, &output);
@@ -4802,6 +4837,205 @@
     return source;
 }
 
+/* static */
+bool AudioPolicyManager::isChannelMaskSpatialized(audio_channel_mask_t channels) {
+    switch (channels) {
+        case AUDIO_CHANNEL_OUT_5POINT1:
+        case AUDIO_CHANNEL_OUT_5POINT1POINT2:
+        case AUDIO_CHANNEL_OUT_5POINT1POINT4:
+        case AUDIO_CHANNEL_OUT_7POINT1:
+        case AUDIO_CHANNEL_OUT_7POINT1POINT2:
+        case AUDIO_CHANNEL_OUT_7POINT1POINT4:
+            return true;
+        default:
+            return false;
+    }
+}
+
+bool AudioPolicyManager::canBeSpatialized(const audio_attributes_t *attr,
+                                      const audio_config_t *config,
+                                      const AudioDeviceTypeAddrVector &devices)  const
+{
+    // The caller can have the audio attributes criteria ignored by either passing a null ptr or
+    // the AUDIO_ATTRIBUTES_INITIALIZER value.
+    // If attributes are specified, current policy is to only allow spatialization for media
+    // and game usages.
+    if (attr != nullptr && *attr != AUDIO_ATTRIBUTES_INITIALIZER) {
+        if (attr->usage != AUDIO_USAGE_MEDIA && attr->usage != AUDIO_USAGE_GAME) {
+            return false;
+        }
+        if ((attr->flags & (AUDIO_FLAG_CONTENT_SPATIALIZED | AUDIO_FLAG_NEVER_SPATIALIZE)) != 0) {
+            return false;
+        }
+    }
+
+    // The caller can have the devices criteria ignored by passing and empty vector, and
+    // getSpatializerOutputProfile() will ignore the devices when looking for a match.
+    // Otherwise an output profile supporting a spatializer effect that can be routed
+    // to the specified devices must exist.
+    sp<IOProfile> profile =
+            getSpatializerOutputProfile(config, devices);
+    if (profile == nullptr) {
+        return false;
+    }
+
+    // The caller can have the audio config criteria ignored by either passing a null ptr or
+    // the AUDIO_CONFIG_INITIALIZER value.
+    // If an audio config is specified, current policy is to only allow spatialization for
+    // some positional channel masks.
+    // If the spatializer output is already opened, only channel masks included in the
+    // spatializer output mixer channel mask are allowed.
+
+    if (config != nullptr && *config != AUDIO_CONFIG_INITIALIZER) {
+        if (!isChannelMaskSpatialized(config->channel_mask)) {
+            return false;
+        }
+        if (mSpatializerOutput != nullptr && mSpatializerOutput->mProfile == profile) {
+            if ((config->channel_mask & mSpatializerOutput->mMixerChannelMask)
+                    != config->channel_mask) {
+                return false;
+            }
+        }
+    }
+    return true;
+}
+
+void AudioPolicyManager::checkVirtualizerClientRoutes() {
+    std::set<audio_stream_type_t> streamsToInvalidate;
+    for (size_t i = 0; i < mOutputs.size(); i++) {
+        const sp<SwAudioOutputDescriptor>& desc = mOutputs[i];
+        for (const sp<TrackClientDescriptor>& client : desc->getClientIterable()) {
+            audio_attributes_t attr = client->attributes();
+            DeviceVector devices = mEngine->getOutputDevicesForAttributes(attr, nullptr, false);
+            AudioDeviceTypeAddrVector devicesTypeAddress = devices.toTypeAddrVector();
+            audio_config_base_t clientConfig = client->config();
+            audio_config_t config = audio_config_initializer(&clientConfig);
+            if (desc != mSpatializerOutput
+                    && canBeSpatialized(&attr, &config, devicesTypeAddress)) {
+                streamsToInvalidate.insert(client->stream());
+            }
+        }
+    }
+
+    for (audio_stream_type_t stream : streamsToInvalidate) {
+        mpClientInterface->invalidateStream(stream);
+    }
+}
+
+status_t AudioPolicyManager::getSpatializerOutput(const audio_config_base_t *mixerConfig,
+                                                        const audio_attributes_t *attr,
+                                                        audio_io_handle_t *output) {
+    *output = AUDIO_IO_HANDLE_NONE;
+
+    DeviceVector devices = mEngine->getOutputDevicesForAttributes(*attr, nullptr, false);
+    AudioDeviceTypeAddrVector devicesTypeAddress = devices.toTypeAddrVector();
+    audio_config_t *configPtr = nullptr;
+    audio_config_t config;
+    if (mixerConfig != nullptr) {
+        config = audio_config_initializer(mixerConfig);
+        configPtr = &config;
+    }
+    if (!canBeSpatialized(attr, configPtr, devicesTypeAddress)) {
+        ALOGW("%s provided attributes or mixer config cannot be spatialized", __func__);
+        return BAD_VALUE;
+    }
+
+    sp<IOProfile> profile =
+            getSpatializerOutputProfile(configPtr, devicesTypeAddress);
+    if (profile == nullptr) {
+        ALOGW("%s no suitable output profile for provided attributes or mixer config", __func__);
+        return BAD_VALUE;
+    }
+
+    if (mSpatializerOutput != nullptr && mSpatializerOutput->mProfile == profile
+            && configPtr != nullptr
+            && configPtr->channel_mask == mSpatializerOutput->mMixerChannelMask) {
+        *output = mSpatializerOutput->mIoHandle;
+        ALOGV("%s returns current spatializer output %d", __func__, *output);
+        return NO_ERROR;
+    }
+    mSpatializerOutput.clear();
+    for (size_t i = 0; i < mOutputs.size(); i++) {
+        sp<SwAudioOutputDescriptor> desc = mOutputs.valueAt(i);
+        if (!desc->isDuplicated() && desc->mProfile == profile) {
+            mSpatializerOutput = desc;
+            break;
+        }
+    }
+    if (mSpatializerOutput == nullptr) {
+        ALOGW("%s no opened spatializer output for profile %s",
+                __func__, profile->getName().c_str());
+        return BAD_VALUE;
+    }
+
+    if (configPtr != nullptr
+            && configPtr->channel_mask != mSpatializerOutput->mMixerChannelMask) {
+        audio_config_base_t savedMixerConfig = {
+            .sample_rate = mSpatializerOutput->getSamplingRate(),
+            .format = mSpatializerOutput->getFormat(),
+            .channel_mask = mSpatializerOutput->mMixerChannelMask,
+        };
+        DeviceVector savedDevices = mSpatializerOutput->devices();
+
+        closeOutput(mSpatializerOutput->mIoHandle);
+        mSpatializerOutput.clear();
+
+        const sp<SwAudioOutputDescriptor> desc =
+                new SwAudioOutputDescriptor(profile, mpClientInterface);
+        status_t status = desc->open(nullptr, mixerConfig, devices,
+                                                    mEngine->getStreamTypeForAttributes(*attr),
+                                                    AUDIO_OUTPUT_FLAG_SPATIALIZER, output);
+        if (status != NO_ERROR) {
+            ALOGW("%s failed opening output: status %d, output %d", __func__, status, *output);
+            if (*output != AUDIO_IO_HANDLE_NONE) {
+                desc->close();
+            }
+            // re open the spatializer output with previous channel mask
+            status_t newStatus = desc->open(nullptr, &savedMixerConfig, savedDevices,
+                                mEngine->getStreamTypeForAttributes(*attr),
+                                AUDIO_OUTPUT_FLAG_SPATIALIZER, output);
+            if (newStatus != NO_ERROR) {
+                if (*output != AUDIO_IO_HANDLE_NONE) {
+                    desc->close();
+                }
+                ALOGE("%s failed to re-open mSpatializerOutput, status %d", __func__, newStatus);
+            } else {
+                mSpatializerOutput = desc;
+                addOutput(*output, desc);
+            }
+            mPreviousOutputs = mOutputs;
+            mpClientInterface->onAudioPortListUpdate();
+            *output = AUDIO_IO_HANDLE_NONE;
+            return status;
+        }
+        mSpatializerOutput = desc;
+        addOutput(*output, desc);
+        mPreviousOutputs = mOutputs;
+        mpClientInterface->onAudioPortListUpdate();
+    }
+
+    checkVirtualizerClientRoutes();
+
+    *output = mSpatializerOutput->mIoHandle;
+    ALOGV("%s returns new spatializer output %d", __func__, *output);
+    return NO_ERROR;
+}
+
+status_t AudioPolicyManager::releaseSpatializerOutput(audio_io_handle_t output) {
+    if (mSpatializerOutput == nullptr) {
+        return INVALID_OPERATION;
+    }
+    if (mSpatializerOutput->mIoHandle != output) {
+        return BAD_VALUE;
+    }
+
+    mSpatializerOutput.clear();
+
+    checkVirtualizerClientRoutes();
+
+    return NO_ERROR;
+}
+
 // ----------------------------------------------------------------------------
 // AudioPolicyManager
 // ----------------------------------------------------------------------------
@@ -4851,6 +5085,8 @@
         ALOGE("could not load audio policy configuration file, setting defaults");
         getConfig().setDefault();
     }
+    //TODO: b/193496180 use spatializer flag at audio HAL when available
+    getConfig().convertSpatializerFlag();
 }
 
 status_t AudioPolicyManager::initialize() {
@@ -4990,7 +5226,8 @@
             sp<SwAudioOutputDescriptor> outputDesc = new SwAudioOutputDescriptor(outProfile,
                                                                                  mpClientInterface);
             audio_io_handle_t output = AUDIO_IO_HANDLE_NONE;
-            status_t status = outputDesc->open(nullptr, DeviceVector(supportedDevice),
+            status_t status = outputDesc->open(nullptr /* halConfig */, nullptr /* mixerConfig */,
+                                               DeviceVector(supportedDevice),
                                                AUDIO_STREAM_DEFAULT,
                                                AUDIO_OUTPUT_FLAG_NONE, &output);
             if (status != NO_ERROR) {
@@ -6995,7 +7232,7 @@
     }
     sp<SwAudioOutputDescriptor> desc = new SwAudioOutputDescriptor(profile, mpClientInterface);
     audio_io_handle_t output = AUDIO_IO_HANDLE_NONE;
-    status_t status = desc->open(nullptr, devices,
+    status_t status = desc->open(nullptr /* halConfig */, nullptr /* mixerConfig */, devices,
             AUDIO_STREAM_DEFAULT, AUDIO_OUTPUT_FLAG_NONE, &output);
     if (status != NO_ERROR) {
         return nullptr;
@@ -7025,7 +7262,7 @@
         config.offload_info.channel_mask = config.channel_mask;
         config.offload_info.format = config.format;
 
-        status = desc->open(&config, devices,
+        status = desc->open(&config, nullptr /* mixerConfig */, devices,
                             AUDIO_STREAM_DEFAULT, AUDIO_OUTPUT_FLAG_NONE, &output);
         if (status != NO_ERROR) {
             return nullptr;
diff --git a/services/audiopolicy/managerdefault/AudioPolicyManager.h b/services/audiopolicy/managerdefault/AudioPolicyManager.h
index 98f96d1..967aa10 100644
--- a/services/audiopolicy/managerdefault/AudioPolicyManager.h
+++ b/services/audiopolicy/managerdefault/AudioPolicyManager.h
@@ -356,6 +356,16 @@
                     BAD_VALUE : NO_ERROR;
         }
 
+        virtual bool canBeSpatialized(const audio_attributes_t *attr,
+                                      const audio_config_t *config,
+                                      const AudioDeviceTypeAddrVector &devices) const;
+
+        virtual status_t getSpatializerOutput(const audio_config_base_t *config,
+                                                const audio_attributes_t *attr,
+                                                audio_io_handle_t *output);
+
+        virtual status_t releaseSpatializerOutput(audio_io_handle_t output);
+
         bool isCallScreenModeSupported() override;
 
         void onNewAudioModulesAvailable() override;
@@ -797,6 +807,8 @@
         sp<SwAudioOutputDescriptor> mPrimaryOutput;     // primary output descriptor
         // list of descriptors for outputs currently opened
 
+        sp<SwAudioOutputDescriptor> mSpatializerOutput;
+
         SwAudioOutputCollection mOutputs;
         // copy of mOutputs before setDeviceConnectionState() opens new outputs
         // reset to mOutputs when updateDevicesAndOutputs() is called.
@@ -933,7 +945,7 @@
         audio_io_handle_t getOutputForDevices(
                 const DeviceVector &devices,
                 audio_session_t session,
-                audio_stream_type_t stream,
+                const audio_attributes_t *attr,
                 const audio_config_t *config,
                 audio_output_flags_t *flags,
                 bool forceMutingHaptic = false);
@@ -948,6 +960,14 @@
                 audio_output_flags_t flags,
                 const DeviceVector &devices,
                 audio_io_handle_t *output);
+
+        sp<IOProfile> getSpatializerOutputProfile(const audio_config_t *config,
+                                                  const AudioDeviceTypeAddrVector &devices) const;
+
+        static bool isChannelMaskSpatialized(audio_channel_mask_t channels);
+
+        void checkVirtualizerClientRoutes();
+
         /**
          * @brief getInputForDevice selects an input handle for a given input device and
          * requester context
diff --git a/services/audiopolicy/service/Android.bp b/services/audiopolicy/service/Android.bp
index 454c020..834478f 100644
--- a/services/audiopolicy/service/Android.bp
+++ b/services/audiopolicy/service/Android.bp
@@ -16,6 +16,8 @@
         "AudioPolicyInterfaceImpl.cpp",
         "AudioPolicyService.cpp",
         "CaptureStateNotifier.cpp",
+        "Spatializer.cpp",
+        "SpatializerPoseController.cpp",
     ],
 
     include_dirs: [
@@ -24,9 +26,11 @@
 
     shared_libs: [
         "libactivitymanager_aidl",
+        "libandroid",
         "libaudioclient",
         "libaudioclient_aidl_conversion",
         "libaudiofoundation",
+        "libaudiohal",
         "libaudiopolicy",
         "libaudiopolicymanagerdefault",
         "libaudioutils",
@@ -34,19 +38,25 @@
         "libcutils",
         "libeffectsconfig",
         "libhardware_legacy",
+        "libheadtracking",
+        "libheadtracking-binding",
         "liblog",
         "libmedia_helper",
         "libmediametrics",
         "libmediautils",
         "libpermission",
+        "libsensor",
         "libsensorprivacy",
+        "libshmemcompat",
         "libutils",
+        "libstagefright_foundation",
         "audioclient-types-aidl-cpp",
         "audioflinger-aidl-cpp",
         "audiopolicy-aidl-cpp",
         "audiopolicy-types-aidl-cpp",
         "capture_state_listener-aidl-cpp",
         "framework-permission-aidl-cpp",
+        "spatializer-aidl-cpp",
     ],
 
     static_libs: [
@@ -55,6 +65,7 @@
     ],
 
     header_libs: [
+        "libaudiohal_headers",
         "libaudiopolicycommon",
         "libaudiopolicyengine_interface_headers",
         "libaudiopolicymanager_interface_headers",
@@ -70,6 +81,8 @@
 
     export_shared_lib_headers: [
         "libactivitymanager_aidl",
+        "libheadtracking",
+        "libheadtracking-binding",
         "libsensorprivacy",
         "framework-permission-aidl-cpp",
     ],
diff --git a/services/audiopolicy/service/AudioPolicyClientImpl.cpp b/services/audiopolicy/service/AudioPolicyClientImpl.cpp
index cd53073..79252d4 100644
--- a/services/audiopolicy/service/AudioPolicyClientImpl.cpp
+++ b/services/audiopolicy/service/AudioPolicyClientImpl.cpp
@@ -40,7 +40,8 @@
 
 status_t AudioPolicyService::AudioPolicyClient::openOutput(audio_module_handle_t module,
                                                            audio_io_handle_t *output,
-                                                           audio_config_t *config,
+                                                           audio_config_t *halConfig,
+                                                           audio_config_base_t *mixerConfig,
                                                            const sp<DeviceDescriptorBase>& device,
                                                            uint32_t *latencyMs,
                                                            audio_output_flags_t flags)
@@ -55,14 +56,17 @@
     media::OpenOutputResponse response;
 
     request.module = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_module_handle_t_int32_t(module));
-    request.config = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_t_AudioConfig(*config));
+    request.halConfig = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_t_AudioConfig(*halConfig));
+    request.mixerConfig =
+            VALUE_OR_RETURN_STATUS(legacy2aidl_audio_config_base_t_AudioConfigBase(*mixerConfig));
     request.device = VALUE_OR_RETURN_STATUS(legacy2aidl_DeviceDescriptorBase(device));
     request.flags = VALUE_OR_RETURN_STATUS(legacy2aidl_audio_output_flags_t_int32_t_mask(flags));
 
     status_t status = af->openOutput(request, &response);
     if (status == OK) {
         *output = VALUE_OR_RETURN_STATUS(aidl2legacy_int32_t_audio_io_handle_t(response.output));
-        *config = VALUE_OR_RETURN_STATUS(aidl2legacy_AudioConfig_audio_config_t(response.config));
+        *halConfig =
+                VALUE_OR_RETURN_STATUS(aidl2legacy_AudioConfig_audio_config_t(response.config));
         *latencyMs = VALUE_OR_RETURN_STATUS(convertIntegral<uint32_t>(response.latencyMs));
     }
     return status;
diff --git a/services/audiopolicy/service/AudioPolicyInterfaceImpl.cpp b/services/audiopolicy/service/AudioPolicyInterfaceImpl.cpp
index 77223b6..4bd1260 100644
--- a/services/audiopolicy/service/AudioPolicyInterfaceImpl.cpp
+++ b/services/audiopolicy/service/AudioPolicyInterfaceImpl.cpp
@@ -121,11 +121,14 @@
     ALOGV("setDeviceConnectionState()");
     Mutex::Autolock _l(mLock);
     AutoCallerClear acc;
-    return binderStatusFromStatusT(
-            mAudioPolicyManager->setDeviceConnectionState(device, state,
+    status_t status = mAudioPolicyManager->setDeviceConnectionState(device, state,
                                                           deviceAidl.address.c_str(),
                                                           deviceNameAidl.c_str(),
-                                                          encodedFormat));
+                                                          encodedFormat);
+    if (status == NO_ERROR) {
+        onCheckSpatializer_l();
+    }
+    return binderStatusFromStatusT(status);
 }
 
 Status AudioPolicyService::getDeviceConnectionState(const media::AudioDevice& deviceAidl,
@@ -165,9 +168,13 @@
     ALOGV("handleDeviceConfigChange()");
     Mutex::Autolock _l(mLock);
     AutoCallerClear acc;
-    return binderStatusFromStatusT(
-            mAudioPolicyManager->handleDeviceConfigChange(device, deviceAidl.address.c_str(),
-                                                          deviceNameAidl.c_str(), encodedFormat));
+    status_t status =  mAudioPolicyManager->handleDeviceConfigChange(
+            device, deviceAidl.address.c_str(), deviceNameAidl.c_str(), encodedFormat);
+
+    if (status == NO_ERROR) {
+       onCheckSpatializer_l();
+    }
+    return binderStatusFromStatusT(status);
 }
 
 Status AudioPolicyService::setPhoneState(media::AudioMode stateAidl, int32_t uidAidl)
@@ -234,6 +241,7 @@
     Mutex::Autolock _l(mLock);
     AutoCallerClear acc;
     mAudioPolicyManager->setForceUse(usage, config);
+    onCheckSpatializer_l();
     return Status::ok();
 }
 
@@ -2062,8 +2070,11 @@
         return binderStatusFromStatusT(NO_INIT);
     }
     Mutex::Autolock _l(mLock);
-    return binderStatusFromStatusT(
-            mAudioPolicyManager->setDevicesRoleForStrategy(strategy, role, devices));
+    status_t status = mAudioPolicyManager->setDevicesRoleForStrategy(strategy, role, devices);
+    if (status == NO_ERROR) {
+       onCheckSpatializer_l();
+    }
+    return binderStatusFromStatusT(status);
 }
 
 Status AudioPolicyService::removeDevicesRoleForStrategy(int32_t strategyAidl,
@@ -2076,8 +2087,11 @@
         return binderStatusFromStatusT(NO_INIT);
     }
     Mutex::Autolock _l(mLock);
-    return binderStatusFromStatusT(
-            mAudioPolicyManager->removeDevicesRoleForStrategy(strategy, role));
+    status_t status = mAudioPolicyManager->removeDevicesRoleForStrategy(strategy, role);
+    if (status == NO_ERROR) {
+       onCheckSpatializer_l();
+    }
+    return binderStatusFromStatusT(status);
 }
 
 Status AudioPolicyService::getDevicesForRoleAndStrategy(
@@ -2205,4 +2219,46 @@
     return Status::ok();
 }
 
+Status AudioPolicyService::getSpatializer(
+        const sp<media::INativeSpatializerCallback>& callback,
+        media::GetSpatializerResponse* _aidl_return) {
+    _aidl_return->spatializer = nullptr;
+    if (callback == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    if (mSpatializer != nullptr) {
+        RETURN_IF_BINDER_ERROR(
+                binderStatusFromStatusT(mSpatializer->registerCallback(callback)));
+        _aidl_return->spatializer = mSpatializer;
+    }
+    return Status::ok();
+}
+
+Status AudioPolicyService::canBeSpatialized(
+        const std::optional<media::AudioAttributesInternal>& attrAidl,
+        const std::optional<media::AudioConfig>& configAidl,
+        const std::vector<media::AudioDevice>& devicesAidl,
+        bool* _aidl_return) {
+    if (mAudioPolicyManager == nullptr) {
+        return binderStatusFromStatusT(NO_INIT);
+    }
+    audio_attributes_t attr = AUDIO_ATTRIBUTES_INITIALIZER;
+    if (attrAidl.has_value()) {
+        attr = VALUE_OR_RETURN_BINDER_STATUS(
+            aidl2legacy_AudioAttributesInternal_audio_attributes_t(attrAidl.value()));
+    }
+    audio_config_t config = AUDIO_CONFIG_INITIALIZER;
+    if (configAidl.has_value()) {
+        config = VALUE_OR_RETURN_BINDER_STATUS(
+                                    aidl2legacy_AudioConfig_audio_config_t(configAidl.value()));
+    }
+    AudioDeviceTypeAddrVector devices = VALUE_OR_RETURN_BINDER_STATUS(
+            convertContainer<AudioDeviceTypeAddrVector>(devicesAidl,
+                                                        aidl2legacy_AudioDeviceTypeAddress));
+
+    Mutex::Autolock _l(mLock);
+    *_aidl_return = mAudioPolicyManager->canBeSpatialized(&attr, &config, devices);
+    return Status::ok();
+}
+
 } // namespace android
diff --git a/services/audiopolicy/service/AudioPolicyService.cpp b/services/audiopolicy/service/AudioPolicyService.cpp
index 4d0e1f1..de71a00 100644
--- a/services/audiopolicy/service/AudioPolicyService.cpp
+++ b/services/audiopolicy/service/AudioPolicyService.cpp
@@ -127,6 +127,7 @@
         loadAudioPolicyManager();
         mAudioPolicyManager = mCreateAudioPolicyManager(mAudioPolicyClient);
     }
+
     // load audio processing modules
     sp<AudioPolicyEffects> audioPolicyEffects = new AudioPolicyEffects();
     sp<UidPolicy> uidPolicy = new UidPolicy(this);
@@ -139,6 +140,15 @@
     }
     uidPolicy->registerSelf();
     sensorPrivacyPolicy->registerSelf();
+
+    // Create spatializer if supported
+    const audio_attributes_t attr = attributes_initializer(AUDIO_USAGE_MEDIA);
+    AudioDeviceTypeAddrVector devices;
+    bool hasSpatializer = mAudioPolicyManager->canBeSpatialized(&attr, nullptr, devices);
+    if (hasSpatializer) {
+        mSpatializer = Spatializer::create(this);
+    }
+    AudioSystem::audioPolicyReady();
 }
 
 void AudioPolicyService::unloadAudioPolicyManager()
@@ -353,6 +363,59 @@
     }
 }
 
+void AudioPolicyService::onCheckSpatializer()
+{
+    Mutex::Autolock _l(mLock);
+    onCheckSpatializer_l();
+}
+
+void AudioPolicyService::onCheckSpatializer_l()
+{
+    if (mSpatializer != nullptr) {
+        mOutputCommandThread->checkSpatializerCommand();
+    }
+}
+
+void AudioPolicyService::doOnCheckSpatializer()
+{
+    Mutex::Autolock _l(mLock);
+
+    if (mSpatializer != nullptr) {
+        if (mSpatializer->getLevel() != media::SpatializationLevel::NONE) {
+            audio_io_handle_t currentOutput = mSpatializer->getOutput();
+            audio_io_handle_t newOutput;
+            const audio_attributes_t attr = attributes_initializer(AUDIO_USAGE_MEDIA);
+            audio_config_base_t config = mSpatializer->getAudioInConfig();
+            status_t status =
+                    mAudioPolicyManager->getSpatializerOutput(&config, &attr, &newOutput);
+
+            if (status == NO_ERROR && currentOutput == newOutput) {
+                return;
+            }
+            mLock.unlock();
+            // It is OK to call detachOutput() is none is already attached.
+            mSpatializer->detachOutput();
+            if (status != NO_ERROR || newOutput == AUDIO_IO_HANDLE_NONE) {
+                mLock.lock();
+                return;
+            }
+            status = mSpatializer->attachOutput(newOutput);
+            mLock.lock();
+            if (status != NO_ERROR) {
+                mAudioPolicyManager->releaseSpatializerOutput(newOutput);
+            }
+        } else if (mSpatializer->getLevel() == media::SpatializationLevel::NONE
+                               && mSpatializer->getOutput() != AUDIO_IO_HANDLE_NONE) {
+            mLock.unlock();
+            audio_io_handle_t output = mSpatializer->detachOutput();
+            mLock.lock();
+            if (output != AUDIO_IO_HANDLE_NONE) {
+                mAudioPolicyManager->releaseSpatializerOutput(output);
+            }
+        }
+    }
+}
+
 status_t AudioPolicyService::clientCreateAudioPatch(const struct audio_patch *patch,
                                                 audio_patch_handle_t *handle,
                                                 int delayMs)
@@ -990,7 +1053,8 @@
         case TRANSACTION_addDevicesRoleForCapturePreset:
         case TRANSACTION_removeDevicesRoleForCapturePreset:
         case TRANSACTION_clearDevicesRoleForCapturePreset:
-        case TRANSACTION_getDevicesForRoleAndCapturePreset: {
+        case TRANSACTION_getDevicesForRoleAndCapturePreset:
+        case TRANSACTION_getSpatializer: {
             if (!isServiceUid(IPCThreadState::self()->getCallingUid())) {
                 ALOGW("%s: transaction %d received from PID %d unauthorized UID %d",
                       __func__, code, IPCThreadState::self()->getCallingPid(),
@@ -1764,6 +1828,17 @@
                     mLock.lock();
                     } break;
 
+                case CHECK_SPATIALIZER: {
+                    ALOGV("AudioCommandThread() processing updateUID states");
+                    svc = mService.promote();
+                    if (svc == 0) {
+                        break;
+                    }
+                    mLock.unlock();
+                    svc->doOnCheckSpatializer();
+                    mLock.lock();
+                    } break;
+
                 default:
                     ALOGW("AudioCommandThread() unknown command %d", command->mCommand);
                 }
@@ -2075,6 +2150,14 @@
     sendCommand(command);
 }
 
+void AudioPolicyService::AudioCommandThread::checkSpatializerCommand()
+{
+    sp<AudioCommand>command = new AudioCommand();
+    command->mCommand = CHECK_SPATIALIZER;
+    ALOGV("AudioCommandThread() adding check spatializer");
+    sendCommand(command);
+}
+
 status_t AudioPolicyService::AudioCommandThread::sendCommand(sp<AudioCommand>& command, int delayMs)
 {
     {
diff --git a/services/audiopolicy/service/AudioPolicyService.h b/services/audiopolicy/service/AudioPolicyService.h
index 7ed829c..27c4e1c 100644
--- a/services/audiopolicy/service/AudioPolicyService.h
+++ b/services/audiopolicy/service/AudioPolicyService.h
@@ -19,6 +19,7 @@
 #define ANDROID_AUDIOPOLICYSERVICE_H
 
 #include <android/media/BnAudioPolicyService.h>
+#include <android/media/GetSpatializerResponse.h>
 #include <android-base/thread_annotations.h>
 #include <cutils/misc.h>
 #include <cutils/config_utils.h>
@@ -38,6 +39,7 @@
 #include <mediautils/ServiceUtilities.h>
 #include "AudioPolicyEffects.h"
 #include "CaptureStateNotifier.h"
+#include "Spatializer.h"
 #include <AudioPolicyInterface.h>
 #include <android/hardware/BnSensorPrivacyListener.h>
 #include <android/content/AttributionSourceState.h>
@@ -53,7 +55,8 @@
 class AudioPolicyService :
     public BinderService<AudioPolicyService>,
     public media::BnAudioPolicyService,
-    public IBinder::DeathRecipient
+    public IBinder::DeathRecipient,
+    public SpatializerPolicyCallback
 {
     friend class BinderService<AudioPolicyService>;
 
@@ -243,11 +246,15 @@
     binder::Status registerSoundTriggerCaptureStateListener(
             const sp<media::ICaptureStateListener>& listener, bool* _aidl_return) override;
 
-    virtual     status_t    onTransact(
-                                uint32_t code,
-                                const Parcel& data,
-                                Parcel* reply,
-                                uint32_t flags);
+    binder::Status getSpatializer(const sp<media::INativeSpatializerCallback>& callback,
+            media::GetSpatializerResponse* _aidl_return) override;
+    binder::Status canBeSpatialized(
+            const std::optional<media::AudioAttributesInternal>& attr,
+            const std::optional<media::AudioConfig>& config,
+            const std::vector<media::AudioDevice>& devices,
+            bool* _aidl_return) override;
+
+    status_t onTransact(uint32_t code, const Parcel& data, Parcel* reply, uint32_t flags) override;
 
     // IBinder::DeathRecipient
     virtual     void        binderDied(const wp<IBinder>& who);
@@ -313,6 +320,16 @@
     void onRoutingUpdated();
     void doOnRoutingUpdated();
 
+    /**
+     * Spatializer SpatializerPolicyCallback implementation.
+     * onCheckSpatializer() sends an event on mOutputCommandThread which executes
+     * doOnCheckSpatializer() to check if a Spatializer output must be opened or closed
+     * by audio policy manager and attach/detach the spatializer effect accordingly.
+     */
+    void onCheckSpatializer() override;
+    void onCheckSpatializer_l();
+    void doOnCheckSpatializer();
+
     void setEffectSuspended(int effectId,
                             audio_session_t sessionId,
                             bool suspended);
@@ -483,7 +500,8 @@
             SET_EFFECT_SUSPENDED,
             AUDIO_MODULES_UPDATE,
             ROUTING_UPDATED,
-            UPDATE_UID_STATES
+            UPDATE_UID_STATES,
+            CHECK_SPATIALIZER
         };
 
         AudioCommandThread (String8 name, const wp<AudioPolicyService>& service);
@@ -532,6 +550,7 @@
                     void        audioModulesUpdateCommand();
                     void        routingChangedCommand();
                     void        updateUidStatesCommand();
+                    void        checkSpatializerCommand();
                     void        insertCommand_l(AudioCommand *command, int delayMs = 0);
     private:
         class AudioCommandData;
@@ -667,7 +686,8 @@
         // The audio policy manager can check if the proposed parameters are suitable or not and act accordingly.
         virtual status_t openOutput(audio_module_handle_t module,
                                     audio_io_handle_t *output,
-                                    audio_config_t *config,
+                                    audio_config_t *halConfig,
+                                    audio_config_base_t *mixerConfig,
                                     const sp<DeviceDescriptorBase>& device,
                                     uint32_t *latencyMs,
                                     audio_output_flags_t flags);
@@ -985,6 +1005,8 @@
 
     CaptureStateNotifier mCaptureStateNotifier;
 
+    sp<Spatializer> mSpatializer;
+
     void *mLibraryHandle = nullptr;
     CreateAudioPolicyManagerInstance mCreateAudioPolicyManager;
     DestroyAudioPolicyManagerInstance mDestroyAudioPolicyManager;
diff --git a/services/audiopolicy/service/Spatializer.cpp b/services/audiopolicy/service/Spatializer.cpp
new file mode 100644
index 0000000..32499f4
--- /dev/null
+++ b/services/audiopolicy/service/Spatializer.cpp
@@ -0,0 +1,753 @@
+/*
+**
+** Copyright 2021, The Android Open Source Project
+**
+** Licensed under the Apache License, Version 2.0 (the "License");
+** you may not use this file except in compliance with the License.
+** You may obtain a copy of the License at
+**
+**     http://www.apache.org/licenses/LICENSE-2.0
+**
+** Unless required by applicable law or agreed to in writing, software
+** distributed under the License is distributed on an "AS IS" BASIS,
+** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+** See the License for the specific language governing permissions and
+** limitations under the License.
+*/
+
+
+#define LOG_TAG "Spatializer"
+//#define LOG_NDEBUG 0
+#include <utils/Log.h>
+
+#include <limits.h>
+#include <stdint.h>
+#include <sys/types.h>
+
+#include <android/content/AttributionSourceState.h>
+#include <android/sensor.h>
+#include <audio_utils/fixedfft.h>
+#include <cutils/bitops.h>
+#include <hardware/sensors.h>
+#include <media/audiohal/EffectsFactoryHalInterface.h>
+#include <media/stagefright/foundation/AHandler.h>
+#include <media/stagefright/foundation/AMessage.h>
+#include <media/ShmemCompat.h>
+#include <mediautils/ServiceUtilities.h>
+#include <utils/Thread.h>
+
+#include "Spatializer.h"
+
+namespace android {
+
+using aidl_utils::statusTFromBinderStatus;
+using aidl_utils::binderStatusFromStatusT;
+using android::content::AttributionSourceState;
+using binder::Status;
+using media::HeadTrackingMode;
+using media::Pose3f;
+using media::SpatializationLevel;
+using media::SpatializationMode;
+using media::SpatializerHeadTrackingMode;
+using media::SensorPoseProvider;
+
+using namespace std::chrono_literals;
+
+#define VALUE_OR_RETURN_BINDER_STATUS(x) \
+    ({ auto _tmp = (x); \
+       if (!_tmp.ok()) return aidl_utils::binderStatusFromStatusT(_tmp.error()); \
+       std::move(_tmp.value()); })
+
+// ---------------------------------------------------------------------------
+
+class Spatializer::EngineCallbackHandler : public AHandler {
+public:
+    EngineCallbackHandler(wp<Spatializer> spatializer)
+            : mSpatializer(spatializer) {
+    }
+
+    enum {
+        // Device state callbacks
+        kWhatOnFramesProcessed,    // AudioEffect::EVENT_FRAMES_PROCESSED
+        kWhatOnHeadToStagePose,    // SpatializerPoseController::Listener::onHeadToStagePose
+        kWhatOnActualModeChange,   // SpatializerPoseController::Listener::onActualModeChange
+    };
+    static constexpr const char *kNumFramesKey = "numFrames";
+    static constexpr const char *kModeKey = "mode";
+    static constexpr const char *kTranslation0Key = "translation0";
+    static constexpr const char *kTranslation1Key = "translation1";
+    static constexpr const char *kTranslation2Key = "translation2";
+    static constexpr const char *kRotation0Key = "rotation0";
+    static constexpr const char *kRotation1Key = "rotation1";
+    static constexpr const char *kRotation2Key = "rotation2";
+
+    void onMessageReceived(const sp<AMessage> &msg) override {
+        switch (msg->what()) {
+            case kWhatOnFramesProcessed: {
+                sp<Spatializer> spatializer = mSpatializer.promote();
+                if (spatializer == nullptr) {
+                    ALOGW("%s: Cannot promote spatializer", __func__);
+                    return;
+                }
+                int numFrames;
+                if (!msg->findInt32(kNumFramesKey, &numFrames)) {
+                    ALOGE("%s: Cannot find num frames!", __func__);
+                    return;
+                }
+                if (numFrames > 0) {
+                    spatializer->calculateHeadPose();
+                }
+                } break;
+            case kWhatOnHeadToStagePose: {
+                sp<Spatializer> spatializer = mSpatializer.promote();
+                if (spatializer == nullptr) {
+                    ALOGW("%s: Cannot promote spatializer", __func__);
+                    return;
+                }
+                std::vector<float> headToStage(sHeadPoseKeys.size());
+                for (size_t i = 0 ; i < sHeadPoseKeys.size(); i++) {
+                    if (!msg->findFloat(sHeadPoseKeys[i], &headToStage[i])) {
+                        ALOGE("%s: Cannot find kTranslation0Key!", __func__);
+                        return;
+                    }
+                }
+                spatializer->onHeadToStagePoseMsg(headToStage);
+                } break;
+            case kWhatOnActualModeChange: {
+                sp<Spatializer> spatializer = mSpatializer.promote();
+                if (spatializer == nullptr) {
+                    ALOGW("%s: Cannot promote spatializer", __func__);
+                    return;
+                }
+                int mode;
+                if (!msg->findInt32(EngineCallbackHandler::kModeKey, &mode)) {
+                    ALOGE("%s: Cannot find actualMode!", __func__);
+                    return;
+                }
+                spatializer->onActualModeChangeMsg(static_cast<HeadTrackingMode>(mode));
+                } break;
+            default:
+                LOG_ALWAYS_FATAL("Invalid callback message %d", msg->what());
+        }
+    }
+private:
+    wp<Spatializer> mSpatializer;
+};
+
+const std::vector<const char *> Spatializer::sHeadPoseKeys = {
+    Spatializer::EngineCallbackHandler::kTranslation0Key,
+    Spatializer::EngineCallbackHandler::kTranslation1Key,
+    Spatializer::EngineCallbackHandler::kTranslation2Key,
+    Spatializer::EngineCallbackHandler::kRotation0Key,
+    Spatializer::EngineCallbackHandler::kRotation1Key,
+    Spatializer::EngineCallbackHandler::kRotation2Key,
+};
+
+// ---------------------------------------------------------------------------
+sp<Spatializer> Spatializer::create(SpatializerPolicyCallback *callback) {
+    sp<Spatializer> spatializer;
+
+    sp<EffectsFactoryHalInterface> effectsFactoryHal = EffectsFactoryHalInterface::create();
+    if (effectsFactoryHal == nullptr) {
+        ALOGW("%s failed to create effect factory interface", __func__);
+        return spatializer;
+    }
+
+    std::vector<effect_descriptor_t> descriptors;
+    status_t status =
+            effectsFactoryHal->getDescriptors(FX_IID_SPATIALIZER, &descriptors);
+    if (status != NO_ERROR) {
+        ALOGW("%s failed to get spatializer descriptor, error %d", __func__, status);
+        return spatializer;
+    }
+    ALOG_ASSERT(!descriptors.empty(),
+            "%s getDescriptors() returned no error but empty list", __func__);
+
+    //TODO: get supported spatialization modes from FX engine or descriptor
+
+    sp<EffectHalInterface> effect;
+    status = effectsFactoryHal->createEffect(&descriptors[0].uuid, AUDIO_SESSION_OUTPUT_STAGE,
+            AUDIO_IO_HANDLE_NONE, AUDIO_PORT_HANDLE_NONE, &effect);
+    ALOGI("%s FX create status %d effect %p", __func__, status, effect.get());
+
+    if (status == NO_ERROR && effect != nullptr) {
+        spatializer = new Spatializer(descriptors[0], callback);
+        if (spatializer->loadEngineConfiguration(effect) != NO_ERROR) {
+            spatializer.clear();
+        }
+    }
+
+    return spatializer;
+}
+
+Spatializer::Spatializer(effect_descriptor_t engineDescriptor, SpatializerPolicyCallback* callback)
+    : mEngineDescriptor(engineDescriptor),
+      mPolicyCallback(callback) {
+    ALOGV("%s", __func__);
+}
+
+void Spatializer::onFirstRef() {
+    mLooper = new ALooper;
+    mLooper->setName("Spatializer-looper");
+    mLooper->start(
+            /*runOnCallingThread*/false,
+            /*canCallJava*/       false,
+            PRIORITY_AUDIO);
+
+    mHandler = new EngineCallbackHandler(this);
+    mLooper->registerHandler(mHandler);
+}
+
+Spatializer::~Spatializer() {
+    ALOGV("%s", __func__);
+    if (mLooper != nullptr) {
+        mLooper->stop();
+        mLooper->unregisterHandler(mHandler->id());
+    }
+    mLooper.clear();
+    mHandler.clear();
+}
+
+status_t Spatializer::loadEngineConfiguration(sp<EffectHalInterface> effect) {
+    ALOGV("%s", __func__);
+
+    std::vector<bool> supportsHeadTracking;
+    status_t status = getHalParameter<false>(effect, SPATIALIZER_PARAM_HEADTRACKING_SUPPORTED,
+                                         &supportsHeadTracking);
+    if (status != NO_ERROR) {
+        return status;
+    }
+    mSupportsHeadTracking = supportsHeadTracking[0];
+
+    status = getHalParameter<true>(effect, SPATIALIZER_PARAM_SUPPORTED_LEVELS, &mLevels);
+    if (status != NO_ERROR) {
+        return status;
+    }
+    status = getHalParameter<true>(effect, SPATIALIZER_PARAM_SUPPORTED_SPATIALIZATION_MODES,
+                                &mSpatializationModes);
+    if (status != NO_ERROR) {
+        return status;
+    }
+    status = getHalParameter<true>(effect, SPATIALIZER_PARAM_SUPPORTED_CHANNEL_MASKS,
+                                 &mChannelMasks);
+    if (status != NO_ERROR) {
+        return status;
+    }
+    return NO_ERROR;
+}
+
+/** Gets the channel mask, sampling rate and format set for the spatializer input. */
+audio_config_base_t Spatializer::getAudioInConfig() const {
+    std::lock_guard lock(mLock);
+    audio_config_base_t config = AUDIO_CONFIG_BASE_INITIALIZER;
+    // For now use highest supported channel count
+    uint32_t maxCount = 0;
+    for ( auto mask : mChannelMasks) {
+        if (audio_channel_count_from_out_mask(mask) > maxCount) {
+            config.channel_mask = mask;
+        }
+    }
+    return config;
+}
+
+status_t Spatializer::registerCallback(
+        const sp<media::INativeSpatializerCallback>& callback) {
+    std::lock_guard lock(mLock);
+    if (callback == nullptr) {
+        return BAD_VALUE;
+    }
+
+    sp<IBinder> binder = IInterface::asBinder(callback);
+    status_t status = binder->linkToDeath(this);
+    if (status == NO_ERROR) {
+        mSpatializerCallback = callback;
+    }
+    ALOGV("%s status %d", __func__, status);
+    return status;
+}
+
+// IBinder::DeathRecipient
+void Spatializer::binderDied(__unused const wp<IBinder> &who) {
+    {
+        std::lock_guard lock(mLock);
+        mLevel = SpatializationLevel::NONE;
+        mSpatializerCallback.clear();
+    }
+    ALOGV("%s", __func__);
+    mPolicyCallback->onCheckSpatializer();
+}
+
+// ISpatializer
+Status Spatializer::getSupportedLevels(std::vector<SpatializationLevel> *levels) {
+    ALOGV("%s", __func__);
+    if (levels == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    levels->push_back(SpatializationLevel::NONE);
+    levels->insert(levels->end(), mLevels.begin(), mLevels.end());
+    return Status::ok();
+}
+
+Status Spatializer::setLevel(SpatializationLevel level) {
+    ALOGV("%s level %d", __func__, (int)level);
+    if (level != SpatializationLevel::NONE
+            && std::find(mLevels.begin(), mLevels.end(), level) == mLevels.end()) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    sp<media::INativeSpatializerCallback> callback;
+    bool levelChanged = false;
+    {
+        std::lock_guard lock(mLock);
+        levelChanged = mLevel != level;
+        mLevel = level;
+        callback = mSpatializerCallback;
+
+        if (levelChanged && mEngine != nullptr) {
+            setEffectParameter_l(SPATIALIZER_PARAM_LEVEL, std::vector<SpatializationLevel>{level});
+        }
+    }
+
+    if (levelChanged) {
+        mPolicyCallback->onCheckSpatializer();
+        if (callback != nullptr) {
+            callback->onLevelChanged(level);
+        }
+    }
+    return Status::ok();
+}
+
+Status Spatializer::getLevel(SpatializationLevel *level) {
+    if (level == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    std::lock_guard lock(mLock);
+    *level = mLevel;
+    ALOGV("%s level %d", __func__, (int)*level);
+    return Status::ok();
+}
+
+Status Spatializer::isHeadTrackingSupported(bool *supports) {
+    ALOGV("%s mSupportsHeadTracking %d", __func__, mSupportsHeadTracking);
+    if (supports == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    std::lock_guard lock(mLock);
+    *supports = mSupportsHeadTracking;
+    return Status::ok();
+}
+
+Status Spatializer::getSupportedHeadTrackingModes(
+        std::vector<SpatializerHeadTrackingMode>* modes) {
+    std::lock_guard lock(mLock);
+    ALOGV("%s", __func__);
+    if (modes == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+
+    modes->push_back(SpatializerHeadTrackingMode::DISABLED);
+    if (mSupportsHeadTracking) {
+        if (mHeadSensor != nullptr) {
+            modes->push_back(SpatializerHeadTrackingMode::RELATIVE_WORLD);
+            if (mScreenSensor != nullptr) {
+                modes->push_back(SpatializerHeadTrackingMode::RELATIVE_SCREEN);
+            }
+        }
+    }
+    return Status::ok();
+}
+
+Status Spatializer::setDesiredHeadTrackingMode(SpatializerHeadTrackingMode mode) {
+    ALOGV("%s mode %d", __func__, (int)mode);
+
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    std::lock_guard lock(mLock);
+    switch (mode) {
+        case SpatializerHeadTrackingMode::OTHER:
+            return binderStatusFromStatusT(BAD_VALUE);
+        case SpatializerHeadTrackingMode::DISABLED:
+            mDesiredHeadTrackingMode = HeadTrackingMode::STATIC;
+            break;
+        case SpatializerHeadTrackingMode::RELATIVE_WORLD:
+            mDesiredHeadTrackingMode = HeadTrackingMode::WORLD_RELATIVE;
+            break;
+        case SpatializerHeadTrackingMode::RELATIVE_SCREEN:
+            mDesiredHeadTrackingMode = HeadTrackingMode::SCREEN_RELATIVE;
+            break;
+    }
+
+    if (mPoseController != nullptr) {
+        mPoseController->setDesiredMode(mDesiredHeadTrackingMode);
+    }
+
+    return Status::ok();
+}
+
+Status Spatializer::getActualHeadTrackingMode(SpatializerHeadTrackingMode *mode) {
+    if (mode == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    std::lock_guard lock(mLock);
+    *mode = mActualHeadTrackingMode;
+    ALOGV("%s mode %d", __func__, (int)*mode);
+    return Status::ok();
+}
+
+Status Spatializer::recenterHeadTracker() {
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    std::lock_guard lock(mLock);
+    if (mPoseController != nullptr) {
+        mPoseController->recenter();
+    }
+    return Status::ok();
+}
+
+Status Spatializer::setGlobalTransform(const std::vector<float>& screenToStage) {
+    ALOGV("%s", __func__);
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    std::optional<Pose3f> maybePose = Pose3f::fromVector(screenToStage);
+    if (!maybePose.has_value()) {
+        ALOGW("Invalid screenToStage vector.");
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    std::lock_guard lock(mLock);
+    if (mPoseController != nullptr) {
+        mPoseController->setScreenToStagePose(maybePose.value());
+    }
+    return Status::ok();
+}
+
+Status Spatializer::release() {
+    ALOGV("%s", __func__);
+    bool levelChanged = false;
+    {
+        std::lock_guard lock(mLock);
+        if (mSpatializerCallback == nullptr) {
+            return binderStatusFromStatusT(INVALID_OPERATION);
+        }
+
+        sp<IBinder> binder = IInterface::asBinder(mSpatializerCallback);
+        binder->unlinkToDeath(this);
+        mSpatializerCallback.clear();
+
+        levelChanged = mLevel != SpatializationLevel::NONE;
+        mLevel = SpatializationLevel::NONE;
+    }
+
+    if (levelChanged) {
+        mPolicyCallback->onCheckSpatializer();
+    }
+    return Status::ok();
+}
+
+Status Spatializer::setHeadSensor(int sensorHandle) {
+    ALOGV("%s sensorHandle %d", __func__, sensorHandle);
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    std::lock_guard lock(mLock);
+    if (sensorHandle == ASENSOR_INVALID) {
+        mHeadSensor = nullptr;
+    } else {
+        mHeadSensor = VALUE_OR_RETURN_BINDER_STATUS(getSensorFromHandle(sensorHandle));
+    }
+    if (mPoseController != nullptr) {
+        mPoseController->setHeadSensor(mHeadSensor);
+    }
+    return Status::ok();
+}
+
+Status Spatializer::setScreenSensor(int sensorHandle) {
+    ALOGV("%s sensorHandle %d", __func__, sensorHandle);
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    std::lock_guard lock(mLock);
+    if (sensorHandle == ASENSOR_INVALID) {
+        mScreenSensor = nullptr;
+    } else {
+        mScreenSensor = VALUE_OR_RETURN_BINDER_STATUS(getSensorFromHandle(sensorHandle));
+    }
+    if (mPoseController != nullptr) {
+        mPoseController->setScreenSensor(mScreenSensor);
+    }
+    return Status::ok();
+}
+
+Status Spatializer::setDisplayOrientation(float physicalToLogicalAngle) {
+    ALOGV("%s physicalToLogicalAngle %f", __func__, physicalToLogicalAngle);
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    std::lock_guard lock(mLock);
+    mDisplayOrientation = physicalToLogicalAngle;
+    if (mPoseController != nullptr) {
+        mPoseController->setDisplayOrientation(mDisplayOrientation);
+    }
+    if (mEngine != nullptr) {
+        setEffectParameter_l(
+            SPATIALIZER_PARAM_DISPLAY_ORIENTATION, std::vector<float>{physicalToLogicalAngle});
+    }
+    return Status::ok();
+}
+
+Status Spatializer::setHingeAngle(float hingeAngle) {
+    std::lock_guard lock(mLock);
+    ALOGV("%s hingeAngle %f", __func__, hingeAngle);
+    if (mEngine != nullptr) {
+        setEffectParameter_l(SPATIALIZER_PARAM_HINGE_ANGLE, std::vector<float>{hingeAngle});
+    }
+    return Status::ok();
+}
+
+Status Spatializer::getSupportedModes(std::vector<SpatializationMode> *modes) {
+    ALOGV("%s", __func__);
+    if (modes == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    *modes = mSpatializationModes;
+    return Status::ok();
+}
+
+Status Spatializer::registerHeadTrackingCallback(
+        const sp<media::ISpatializerHeadTrackingCallback>& callback) {
+    ALOGV("%s callback %p", __func__, callback.get());
+    std::lock_guard lock(mLock);
+    if (!mSupportsHeadTracking) {
+        return binderStatusFromStatusT(INVALID_OPERATION);
+    }
+    mHeadTrackingCallback = callback;
+    return Status::ok();
+}
+
+Status Spatializer::setParameter(int key, const std::vector<unsigned char>& value) {
+    ALOGV("%s key %d", __func__, key);
+    std::lock_guard lock(mLock);
+    status_t status = INVALID_OPERATION;
+    if (mEngine != nullptr) {
+        status = setEffectParameter_l(key, value);
+    }
+    return binderStatusFromStatusT(status);
+}
+
+Status Spatializer::getParameter(int key, std::vector<unsigned char> *value) {
+    ALOGV("%s key %d value size %d", __func__, key,
+          (value != nullptr ? (int)value->size() : -1));
+    if (value == nullptr) {
+        return binderStatusFromStatusT(BAD_VALUE);
+    }
+    std::lock_guard lock(mLock);
+    status_t status = INVALID_OPERATION;
+    if (mEngine != nullptr) {
+        ALOGV("%s key %d mEngine %p", __func__, key, mEngine.get());
+        status = getEffectParameter_l(key, value);
+    }
+    return binderStatusFromStatusT(status);
+}
+
+Status Spatializer::getOutput(int *output) {
+    ALOGV("%s", __func__);
+    if (output == nullptr) {
+        binderStatusFromStatusT(BAD_VALUE);
+    }
+    std::lock_guard lock(mLock);
+    *output = VALUE_OR_RETURN_BINDER_STATUS(legacy2aidl_audio_io_handle_t_int32_t(mOutput));
+    ALOGV("%s got output %d", __func__, *output);
+    return Status::ok();
+}
+
+// SpatializerPoseController::Listener
+void Spatializer::onHeadToStagePose(const Pose3f& headToStage) {
+    ALOGV("%s", __func__);
+    LOG_ALWAYS_FATAL_IF(!mSupportsHeadTracking,
+            "onHeadToStagePose() called with no head tracking support!");
+
+    auto vec = headToStage.toVector();
+    LOG_ALWAYS_FATAL_IF(vec.size() != sHeadPoseKeys.size(),
+            "%s invalid head to stage vector size %zu", __func__, vec.size());
+
+    sp<AMessage> msg =
+            new AMessage(EngineCallbackHandler::kWhatOnHeadToStagePose, mHandler);
+    for (size_t i = 0 ; i < sHeadPoseKeys.size(); i++) {
+        msg->setFloat(sHeadPoseKeys[i], vec[i]);
+    }
+    msg->post();
+}
+
+void Spatializer::onHeadToStagePoseMsg(const std::vector<float>& headToStage) {
+    ALOGV("%s", __func__);
+    sp<media::ISpatializerHeadTrackingCallback> callback;
+    {
+        std::lock_guard lock(mLock);
+        callback = mHeadTrackingCallback;
+        if (mEngine != nullptr) {
+            setEffectParameter_l(SPATIALIZER_PARAM_HEAD_TO_STAGE, headToStage);
+        }
+    }
+
+    if (callback != nullptr) {
+        callback->onHeadToSoundStagePoseUpdated(headToStage);
+    }
+}
+
+void Spatializer::onActualModeChange(HeadTrackingMode mode) {
+    ALOGV("%s(%d)", __func__, (int)mode);
+    sp<AMessage> msg =
+            new AMessage(EngineCallbackHandler::kWhatOnActualModeChange, mHandler);
+    msg->setInt32(EngineCallbackHandler::kModeKey, static_cast<int>(mode));
+    msg->post();
+}
+
+void Spatializer::onActualModeChangeMsg(HeadTrackingMode mode) {
+    ALOGV("%s(%d)", __func__, (int) mode);
+    sp<media::ISpatializerHeadTrackingCallback> callback;
+    SpatializerHeadTrackingMode spatializerMode;
+    {
+        std::lock_guard lock(mLock);
+        if (!mSupportsHeadTracking) {
+            spatializerMode = SpatializerHeadTrackingMode::DISABLED;
+        } else {
+            switch (mode) {
+                case HeadTrackingMode::STATIC:
+                    spatializerMode = SpatializerHeadTrackingMode::DISABLED;
+                    break;
+                case HeadTrackingMode::WORLD_RELATIVE:
+                    spatializerMode = SpatializerHeadTrackingMode::RELATIVE_WORLD;
+                    break;
+                case HeadTrackingMode::SCREEN_RELATIVE:
+                    spatializerMode = SpatializerHeadTrackingMode::RELATIVE_SCREEN;
+                    break;
+                default:
+                    LOG_ALWAYS_FATAL("Unknown mode: %d", mode);
+            }
+        }
+        mActualHeadTrackingMode = spatializerMode;
+        callback = mHeadTrackingCallback;
+    }
+    if (callback != nullptr) {
+        callback->onHeadTrackingModeChanged(spatializerMode);
+    }
+}
+
+/* static */
+ConversionResult<ASensorRef> Spatializer::getSensorFromHandle(int handle) {
+    ASensorManager* sensorManager =
+            ASensorManager_getInstanceForPackage("headtracker");
+    if (!sensorManager) {
+        ALOGE("Failed to get a sensor manager");
+        return base::unexpected(NO_INIT);
+    }
+    ASensorList sensorList;
+    int numSensors = ASensorManager_getSensorList(sensorManager, &sensorList);
+    for (int i = 0; i < numSensors; ++i) {
+        if (ASensor_getHandle(sensorList[i]) == handle) {
+            return sensorList[i];
+        }
+    }
+    return base::unexpected(BAD_VALUE);
+}
+
+status_t Spatializer::attachOutput(audio_io_handle_t output) {
+    std::shared_ptr<SpatializerPoseController> poseController;
+    {
+        std::lock_guard lock(mLock);
+        ALOGV("%s output %d mOutput %d", __func__, (int)output, (int)mOutput);
+        if (mOutput != AUDIO_IO_HANDLE_NONE) {
+            LOG_ALWAYS_FATAL_IF(mEngine == nullptr, "%s output set without FX engine", __func__);
+            // remove FX instance
+            mEngine->setEnabled(false);
+            mEngine.clear();
+        }
+        // create FX instance on output
+        AttributionSourceState attributionSource = AttributionSourceState();
+        mEngine = new AudioEffect(attributionSource);
+        mEngine->set(nullptr, &mEngineDescriptor.uuid, 0, Spatializer::engineCallback /* cbf */,
+                     this /* user */, AUDIO_SESSION_OUTPUT_STAGE, output, {} /* device */,
+                     false /* probe */, true /* notifyFramesProcessed */);
+        status_t status = mEngine->initCheck();
+        ALOGV("%s mEngine create status %d", __func__, (int)status);
+        if (status != NO_ERROR) {
+            return status;
+        }
+
+        setEffectParameter_l(SPATIALIZER_PARAM_LEVEL,
+                             std::vector<SpatializationLevel>{mLevel});
+        setEffectParameter_l(SPATIALIZER_PARAM_HEADTRACKING_MODE,
+                             std::vector<SpatializerHeadTrackingMode>{mActualHeadTrackingMode});
+
+        mEngine->setEnabled(true);
+        mOutput = output;
+
+        if (mSupportsHeadTracking) {
+            mPoseController = std::make_shared<SpatializerPoseController>(
+                    static_cast<SpatializerPoseController::Listener*>(this), 10ms, 50ms);
+            LOG_ALWAYS_FATAL_IF(mPoseController == nullptr,
+                                "%s could not allocate pose controller", __func__);
+
+            mPoseController->setDesiredMode(mDesiredHeadTrackingMode);
+            mPoseController->setHeadSensor(mHeadSensor);
+            mPoseController->setScreenSensor(mScreenSensor);
+            mPoseController->setDisplayOrientation(mDisplayOrientation);
+            poseController = mPoseController;
+        }
+    }
+    if (poseController != nullptr) {
+        poseController->waitUntilCalculated();
+    }
+    return NO_ERROR;
+}
+
+audio_io_handle_t Spatializer::detachOutput() {
+    std::lock_guard lock(mLock);
+    ALOGV("%s mOutput %d", __func__, (int)mOutput);
+    audio_io_handle_t output = AUDIO_IO_HANDLE_NONE;
+    if (mOutput == AUDIO_IO_HANDLE_NONE) {
+        return output;
+    }
+    // remove FX instance
+    mEngine->setEnabled(false);
+    mEngine.clear();
+    output = mOutput;
+    mOutput = AUDIO_IO_HANDLE_NONE;
+    mPoseController.reset();
+    return output;
+}
+
+void Spatializer::calculateHeadPose() {
+    ALOGV("%s", __func__);
+    std::lock_guard lock(mLock);
+    if (mPoseController != nullptr) {
+        mPoseController->calculateAsync();
+    }
+}
+
+void Spatializer::engineCallback(int32_t event, void *user, void *info) {
+    if (user == nullptr) {
+        return;
+    }
+    Spatializer* const me = reinterpret_cast<Spatializer *>(user);
+    switch (event) {
+        case AudioEffect::EVENT_FRAMES_PROCESSED: {
+            int frames = info == nullptr ? 0 : *(int*)info;
+            ALOGD("%s frames processed %d for me %p", __func__, frames, me);
+            me->postFramesProcessedMsg(frames);
+        } break;
+        default:
+            ALOGD("%s event %d", __func__, event);
+            break;
+    }
+}
+
+void Spatializer::postFramesProcessedMsg(int frames) {
+    sp<AMessage> msg =
+            new AMessage(EngineCallbackHandler::kWhatOnFramesProcessed, mHandler);
+    msg->setInt32(EngineCallbackHandler::kNumFramesKey, frames);
+    msg->post();
+}
+
+} // namespace android
diff --git a/services/audiopolicy/service/Spatializer.h b/services/audiopolicy/service/Spatializer.h
new file mode 100644
index 0000000..136a467
--- /dev/null
+++ b/services/audiopolicy/service/Spatializer.h
@@ -0,0 +1,341 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef ANDROID_MEDIA_SPATIALIZER_H
+#define ANDROID_MEDIA_SPATIALIZER_H
+
+#include <android/media/BnEffect.h>
+#include <android/media/BnSpatializer.h>
+#include <android/media/SpatializationLevel.h>
+#include <android/media/SpatializationMode.h>
+#include <android/media/SpatializerHeadTrackingMode.h>
+#include <android/sensor.h>
+#include <media/audiohal/EffectHalInterface.h>
+#include <media/stagefright/foundation/ALooper.h>
+#include <media/AudioEffect.h>
+#include <system/audio_effects/effect_spatializer.h>
+
+#include "SpatializerPoseController.h"
+
+namespace android {
+
+
+// ----------------------------------------------------------------------------
+
+/**
+ * A callback interface from the Spatializer object or its parent AudioPolicyService.
+ * This is implemented by the audio policy service hosting the Spatializer to perform
+ * actions needed when a state change inside the Spatializer requires some audio system
+ * changes that cannot be performed by the Spatializer. For instance opening or closing a
+ * spatializer output stream when the spatializer is enabled or disabled
+ */
+class SpatializerPolicyCallback {
+public:
+    /** Called when a stage change occurs that requires the parent audio policy service to take
+     * some action.
+     */
+    virtual void onCheckSpatializer() = 0;
+
+    virtual ~SpatializerPolicyCallback() = default;
+};
+/**
+ * The Spatializer class implements all functional controlling the multichannel spatializer
+ * with head tracking implementation in the native audio service: audio policy and audio flinger.
+ * It presents an AIDL interface available to the java audio service to discover the availability
+ * of the feature and options, control its state and register an active head tracking sensor.
+ * It maintains the current state of the platform spatializer and applies the stored parameters
+ * when the spatializer engine is created and enabled.
+ * Based on the requested spatializer level, it will request the creation of a specialized output
+ * mixer to the audio policy service which will in turn notify the Spatializer of the output
+ * stream on which a spatializer engine should be created, configured and enabled.
+ * The spatializer also hosts the head tracking management logic. This logic receives the
+ * desired head tracking mode and selected head tracking sensor, registers a sensor event listener
+ * and derives the compounded head pose information to the spatializer engine.
+ *
+ * Workflow:
+ * - Initialization: when the audio policy service starts, it checks if a spatializer effect
+ * engine exists and if the audio policy manager reports a dedicated spatializer output profile.
+ * If both conditions are met, a Spatializer object is created
+ * - Capabilities discovery: AudioService will call AudioSystem::canBeSpatialized() and if true,
+ * acquire an ISpatializer interface with AudioSystem::getSpatializer(). This interface
+ * will be used to query the implementation capabilities and configure the spatializer.
+ * - Enabling: when ISpatializer::setLevel() sets a level different from NONE the spatializer
+ * is considered enabled. The audio policy callback onCheckSpatializer() is called. This
+ * triggers a request to audio policy manager to open a spatialization output stream and a
+ * spatializer mixer is created in audio flinger. When an output is returned by audio policy
+ * manager, Spatializer::attachOutput() is called which creates and enables the spatializer
+ * stage engine on the specified output.
+ * - Disabling: when the spatialization level is set to NONE, the spatializer is considered
+ * disabled. The audio policy callback onCheckSpatializer() is called. This triggers a call
+ * to Spatializer::detachOutput() and the spatializer engine is released. Then a request is
+ * made to audio policy manager to release and close the spatializer output stream and the
+ * spatializer mixer thread is destroyed.
+ */
+class Spatializer : public media::BnSpatializer,
+                    public IBinder::DeathRecipient,
+                    private SpatializerPoseController::Listener {
+  public:
+    static sp<Spatializer> create(SpatializerPolicyCallback *callback);
+
+           ~Spatializer() override;
+
+    /** RefBase */
+    void onFirstRef();
+
+    /** ISpatializer, see ISpatializer.aidl */
+    binder::Status release() override;
+    binder::Status getSupportedLevels(std::vector<media::SpatializationLevel>* levels) override;
+    binder::Status setLevel(media::SpatializationLevel level) override;
+    binder::Status getLevel(media::SpatializationLevel *level) override;
+    binder::Status isHeadTrackingSupported(bool *supports);
+    binder::Status getSupportedHeadTrackingModes(
+            std::vector<media::SpatializerHeadTrackingMode>* modes) override;
+    binder::Status setDesiredHeadTrackingMode(
+            media::SpatializerHeadTrackingMode mode) override;
+    binder::Status getActualHeadTrackingMode(
+            media::SpatializerHeadTrackingMode* mode) override;
+    binder::Status recenterHeadTracker() override;
+    binder::Status setGlobalTransform(const std::vector<float>& screenToStage) override;
+    binder::Status setHeadSensor(int sensorHandle) override;
+    binder::Status setScreenSensor(int sensorHandle) override;
+    binder::Status setDisplayOrientation(float physicalToLogicalAngle) override;
+    binder::Status setHingeAngle(float hingeAngle) override;
+    binder::Status getSupportedModes(std::vector<media::SpatializationMode>* modes) override;
+    binder::Status registerHeadTrackingCallback(
+        const sp<media::ISpatializerHeadTrackingCallback>& callback) override;
+    binder::Status setParameter(int key, const std::vector<unsigned char>& value) override;
+    binder::Status getParameter(int key, std::vector<unsigned char> *value) override;
+    binder::Status getOutput(int *output);
+
+    /** IBinder::DeathRecipient. Listen to the death of the INativeSpatializerCallback. */
+    virtual void binderDied(const wp<IBinder>& who);
+
+    /** Registers a INativeSpatializerCallback when a client is attached to this Spatializer
+     * by audio policy service.
+     */
+    status_t registerCallback(const sp<media::INativeSpatializerCallback>& callback);
+
+    status_t loadEngineConfiguration(sp<EffectHalInterface> effect);
+
+    /** Level getter for use by local classes. */
+    media::SpatializationLevel getLevel() const { std::lock_guard lock(mLock); return mLevel; }
+
+    /** Called by audio policy service when the special output mixer dedicated to spatialization
+     * is opened and the spatializer engine must be created.
+     */
+    status_t attachOutput(audio_io_handle_t output);
+    /** Called by audio policy service when the special output mixer dedicated to spatialization
+     * is closed and the spatializer engine must be release.
+     */
+    audio_io_handle_t detachOutput();
+    /** Returns the output stream the spatializer is attached to. */
+    audio_io_handle_t getOutput() const { std::lock_guard lock(mLock); return mOutput; }
+
+    /** Gets the channel mask, sampling rate and format set for the spatializer input. */
+    audio_config_base_t getAudioInConfig() const;
+
+    void calculateHeadPose();
+
+private:
+    Spatializer(effect_descriptor_t engineDescriptor,
+                     SpatializerPolicyCallback *callback);
+
+    static void engineCallback(int32_t event, void* user, void *info);
+
+    // From VirtualizerStageController::Listener
+    void onHeadToStagePose(const media::Pose3f& headToStage) override;
+    void onActualModeChange(media::HeadTrackingMode mode) override;
+
+    void onHeadToStagePoseMsg(const std::vector<float>& headToStage);
+    void onActualModeChangeMsg(media::HeadTrackingMode mode);
+
+
+    static ConversionResult<ASensorRef> getSensorFromHandle(int handle);
+
+    static constexpr int kMaxEffectParamValues = 10;
+    /**
+     * Get a parameter from spatializer engine by calling the effect HAL command method directly.
+     * To be used when the engine instance mEngine is not yet created in the effect framework.
+     * When MULTI_VALUES is false, the expected reply is only one value of type T.
+     * When MULTI_VALUES is true, the expected reply is made of a number (of type T) indicating
+     * how many values are returned, followed by this number for values of type T.
+     */
+    template<bool MULTI_VALUES, typename T>
+    status_t getHalParameter(sp<EffectHalInterface> effect, uint32_t type,
+                                          std::vector<T> *values) {
+        static_assert(sizeof(T) <= sizeof(uint32_t), "The size of T must less than 32 bits");
+
+        uint32_t cmd[sizeof(effect_param_t) / sizeof(uint32_t) + 1];
+        uint32_t reply[sizeof(effect_param_t) / sizeof(uint32_t) + 2 + kMaxEffectParamValues];
+
+        effect_param_t *p = (effect_param_t *)cmd;
+        p->psize = sizeof(uint32_t);
+        if (MULTI_VALUES) {
+            p->vsize = (kMaxEffectParamValues + 1) * sizeof(T);
+        } else {
+            p->vsize = sizeof(T);
+        }
+        *(uint32_t *)p->data = type;
+        uint32_t replySize = sizeof(effect_param_t) + p->psize + p->vsize;
+
+        status_t status = effect->command(EFFECT_CMD_GET_PARAM,
+                                          sizeof(effect_param_t) + sizeof(uint32_t), cmd,
+                                          &replySize, reply);
+        if (status != NO_ERROR) {
+            return status;
+        }
+        if (p->status != NO_ERROR) {
+            return p->status;
+        }
+        if (replySize <
+                sizeof(effect_param_t) + sizeof(uint32_t) + (MULTI_VALUES ? 2 : 1) * sizeof(T)) {
+            return BAD_VALUE;
+        }
+
+        T *params = (T *)((uint8_t *)reply + sizeof(effect_param_t) + sizeof(uint32_t));
+        int numParams = 1;
+        if (MULTI_VALUES) {
+            numParams = (int)*params++;
+        }
+        if (numParams > kMaxEffectParamValues) {
+            return BAD_VALUE;
+        }
+        (*values).clear();
+        std::copy(&params[0], &params[numParams], back_inserter(*values));
+        return NO_ERROR;
+    }
+
+    /**
+     * Set a parameter to spatializer engine by calling setParameter on mEngine AudioEffect object.
+     * It is possible to pass more than one value of type T according to the parameter type
+     *  according to values vector size.
+     */
+    template<typename T>
+    status_t setEffectParameter_l(uint32_t type, const std::vector<T>& values) REQUIRES(mLock) {
+        static_assert(sizeof(T) <= sizeof(uint32_t), "The size of T must less than 32 bits");
+
+        uint32_t cmd[sizeof(effect_param_t) / sizeof(uint32_t) + 1 + values.size()];
+        effect_param_t *p = (effect_param_t *)cmd;
+        p->psize = sizeof(uint32_t);
+        p->vsize = sizeof(T) * values.size();
+        *(uint32_t *)p->data = type;
+        memcpy((uint32_t *)p->data + 1, values.data(), sizeof(T) * values.size());
+
+        status_t status = mEngine->setParameter(p);
+        if (status != NO_ERROR) {
+            return status;
+        }
+        if (p->status != NO_ERROR) {
+            return p->status;
+        }
+        return NO_ERROR;
+    }
+
+    /**
+     * Get a parameter from spatializer engine by calling getParameter on AudioEffect object.
+     * It is possible to read more than one value of type T according to the parameter type
+     * by specifying values vector size.
+     */
+    template<typename T>
+    status_t getEffectParameter_l(uint32_t type, std::vector<T> *values) REQUIRES(mLock) {
+        static_assert(sizeof(T) <= sizeof(uint32_t), "The size of T must less than 32 bits");
+
+        uint32_t cmd[sizeof(effect_param_t) / sizeof(uint32_t) + 1 + values->size()];
+        effect_param_t *p = (effect_param_t *)cmd;
+        p->psize = sizeof(uint32_t);
+        p->vsize = sizeof(T) * values->size();
+        *(uint32_t *)p->data = type;
+
+        status_t status = mEngine->getParameter(p);
+
+        if (status != NO_ERROR) {
+            return status;
+        }
+        if (p->status != NO_ERROR) {
+            return p->status;
+        }
+
+        int numValues = std::min(p->vsize / sizeof(T), values->size());
+        (*values).clear();
+        T *retValues = (T *)((uint8_t *)p->data + sizeof(uint32_t));
+        std::copy(&retValues[0], &retValues[numValues], back_inserter(*values));
+
+        return NO_ERROR;
+    }
+
+    void postFramesProcessedMsg(int frames);
+
+    /** Effect engine descriptor */
+    const effect_descriptor_t mEngineDescriptor;
+    /** Callback interface to parent audio policy service */
+    SpatializerPolicyCallback* mPolicyCallback;
+
+    /** Mutex protecting internal state */
+    mutable std::mutex mLock;
+
+    /** Client AudioEffect for the engine */
+    sp<AudioEffect> mEngine GUARDED_BY(mLock);
+    /** Output stream the spatializer mixer thread is attached to */
+    audio_io_handle_t mOutput GUARDED_BY(mLock) = AUDIO_IO_HANDLE_NONE;
+
+    /** Callback interface to the client (AudioService) controlling this`Spatializer */
+    sp<media::INativeSpatializerCallback> mSpatializerCallback GUARDED_BY(mLock);
+
+    /** Callback interface for head tracking */
+    sp<media::ISpatializerHeadTrackingCallback> mHeadTrackingCallback GUARDED_BY(mLock);
+
+    /** Requested spatialization level */
+    media::SpatializationLevel mLevel GUARDED_BY(mLock) = media::SpatializationLevel::NONE;
+
+    /** Control logic for head-tracking, etc. */
+    std::shared_ptr<SpatializerPoseController> mPoseController GUARDED_BY(mLock);
+
+    /** Last requested head tracking mode */
+    media::HeadTrackingMode mDesiredHeadTrackingMode GUARDED_BY(mLock)
+            = media::HeadTrackingMode::STATIC;
+
+    /** Last-reported actual head-tracking mode. */
+    media::SpatializerHeadTrackingMode mActualHeadTrackingMode GUARDED_BY(mLock)
+            = media::SpatializerHeadTrackingMode::DISABLED;
+
+    /** Selected Head pose sensor */
+    ASensorRef mHeadSensor GUARDED_BY(mLock) = nullptr;
+
+    /** Selected Screen pose sensor */
+    ASensorRef mScreenSensor GUARDED_BY(mLock) = nullptr;
+
+    /** Last display orientation received */
+    static constexpr float kDisplayOrientationInvalid = 1000;
+    float mDisplayOrientation GUARDED_BY(mLock) = kDisplayOrientationInvalid;
+
+    std::vector<media::SpatializationLevel> mLevels;
+    std::vector<media::SpatializationMode> mSpatializationModes;
+    std::vector<audio_channel_mask_t> mChannelMasks;
+    bool mSupportsHeadTracking;
+
+    // Looper thread for mEngine callbacks
+    class EngineCallbackHandler;
+
+    sp<ALooper> mLooper;
+    sp<EngineCallbackHandler> mHandler;
+
+    static const std::vector<const char *> sHeadPoseKeys;
+};
+
+
+}; // namespace android
+
+#endif // ANDROID_MEDIA_SPATIALIZER_H
diff --git a/services/audiopolicy/service/SpatializerPoseController.cpp b/services/audiopolicy/service/SpatializerPoseController.cpp
new file mode 100644
index 0000000..215f899
--- /dev/null
+++ b/services/audiopolicy/service/SpatializerPoseController.cpp
@@ -0,0 +1,234 @@
+/*
+ * Copyright 2021, The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#include "SpatializerPoseController.h"
+
+#define LOG_TAG "SpatializerPoseController"
+//#define LOG_NDEBUG 0
+#include <utils/Log.h>
+#include <utils/SystemClock.h>
+
+namespace android {
+
+using media::createHeadTrackingProcessor;
+using media::HeadTrackingMode;
+using media::HeadTrackingProcessor;
+using media::Pose3f;
+using media::SensorPoseProvider;
+using media::Twist3f;
+
+using namespace std::chrono_literals;
+
+namespace {
+
+// This is how fast, in m/s, we allow position to shift during rate-limiting.
+constexpr auto kMaxTranslationalVelocity = 2;
+
+// This is how fast, in rad/s, we allow rotation angle to shift during rate-limiting.
+constexpr auto kMaxRotationalVelocity = 4 * M_PI;
+
+// This should be set to the typical time scale that the translation sensors used drift in. This
+// means, loosely, for how long we can trust the reading to be "accurate enough". This would
+// determine the time constants used for high-pass filtering those readings. If the value is set
+// too high, we may experience drift. If it is set too low, we may experience poses tending toward
+// identity too fast.
+constexpr auto kTranslationalDriftTimeConstant = 20s;
+
+// This should be set to the typical time scale that the rotation sensors used drift in. This
+// means, loosely, for how long we can trust the reading to be "accurate enough". This would
+// determine the time constants used for high-pass filtering those readings. If the value is set
+// too high, we may experience drift. If it is set too low, we may experience poses tending toward
+// identity too fast.
+constexpr auto kRotationalDriftTimeConstant = 20s;
+
+// This is how far into the future we predict the head pose, using linear extrapolation based on
+// twist (velocity). It should be set to a value that matches the characteristic durations of moving
+// one's head. The higher we set this, the more latency we are able to reduce, but setting this too
+// high will result in high prediction errors whenever the head accelerates (changes velocity).
+constexpr auto kPredictionDuration = 10ms;
+
+// After losing this many consecutive samples from either sensor, we would treat the measurement as
+// stale;
+constexpr auto kMaxLostSamples = 4;
+
+// Time units for system clock ticks. This is what the Sensor Framework timestamps represent and
+// what we use for pose filtering.
+using Ticks = std::chrono::nanoseconds;
+
+// How many ticks in a second.
+constexpr auto kTicksPerSecond = Ticks::period::den;
+
+}  // namespace
+
+SpatializerPoseController::SpatializerPoseController(Listener* listener,
+                                                     std::chrono::microseconds sensorPeriod,
+                                                     std::chrono::microseconds maxUpdatePeriod)
+    : mListener(listener),
+      mSensorPeriod(sensorPeriod),
+      mProcessor(createHeadTrackingProcessor(HeadTrackingProcessor::Options{
+              .maxTranslationalVelocity = kMaxTranslationalVelocity / kTicksPerSecond,
+              .maxRotationalVelocity = kMaxRotationalVelocity / kTicksPerSecond,
+              .translationalDriftTimeConstant = Ticks(kTranslationalDriftTimeConstant).count(),
+              .rotationalDriftTimeConstant = Ticks(kRotationalDriftTimeConstant).count(),
+              .freshnessTimeout = Ticks(sensorPeriod * kMaxLostSamples).count(),
+              .predictionDuration = Ticks(kPredictionDuration).count(),
+      })),
+      mPoseProvider(SensorPoseProvider::create("headtracker", this)),
+      mThread([this, maxUpdatePeriod] {
+          while (true) {
+              Pose3f headToStage;
+              std::optional<HeadTrackingMode> modeIfChanged;
+              {
+                  std::unique_lock lock(mMutex);
+                  mCondVar.wait_for(lock, maxUpdatePeriod,
+                                    [this] { return mShouldExit || mShouldCalculate; });
+                  if (mShouldExit) {
+                      ALOGV("Exiting thread");
+                      return;
+                  }
+
+                  // Calculate.
+                  std::tie(headToStage, modeIfChanged) = calculate_l();
+              }
+
+              // Invoke the callbacks outside the lock.
+              mListener->onHeadToStagePose(headToStage);
+              if (modeIfChanged) {
+                  mListener->onActualModeChange(modeIfChanged.value());
+              }
+
+              {
+                  std::lock_guard lock(mMutex);
+                  if (!mCalculated) {
+                      mCalculated = true;
+                      mCondVar.notify_all();
+                  }
+                  mShouldCalculate = false;
+              }
+          }
+      }) {}
+
+SpatializerPoseController::~SpatializerPoseController() {
+    {
+        std::unique_lock lock(mMutex);
+        mShouldExit = true;
+        mCondVar.notify_all();
+    }
+    mThread.join();
+}
+
+void SpatializerPoseController::setHeadSensor(const ASensor* sensor) {
+    std::lock_guard lock(mMutex);
+    // Stop current sensor, if valid and different from the other sensor.
+    if (mHeadSensor != SensorPoseProvider::INVALID_HANDLE && mHeadSensor != mScreenSensor) {
+        mPoseProvider->stopSensor(mHeadSensor);
+    }
+
+    if (sensor != nullptr) {
+        if (ASensor_getHandle(sensor) != mScreenSensor) {
+            // Start new sensor.
+            mHeadSensor = mPoseProvider->startSensor(sensor, mSensorPeriod);
+        } else {
+            // Sensor is already enabled.
+            mHeadSensor = mScreenSensor;
+        }
+    } else {
+        mHeadSensor = SensorPoseProvider::INVALID_HANDLE;
+    }
+
+    mProcessor->recenter(true, false);
+}
+
+void SpatializerPoseController::setScreenSensor(const ASensor* sensor) {
+    std::lock_guard lock(mMutex);
+    // Stop current sensor, if valid and different from the other sensor.
+    if (mScreenSensor != SensorPoseProvider::INVALID_HANDLE && mScreenSensor != mHeadSensor) {
+        mPoseProvider->stopSensor(mScreenSensor);
+    }
+
+    if (sensor != nullptr) {
+        if (ASensor_getHandle(sensor) != mHeadSensor) {
+            // Start new sensor.
+            mScreenSensor = mPoseProvider->startSensor(sensor, mSensorPeriod);
+        } else {
+            // Sensor is already enabled.
+            mScreenSensor = mHeadSensor;
+        }
+    } else {
+        mScreenSensor = SensorPoseProvider::INVALID_HANDLE;
+    }
+
+    mProcessor->recenter(false, true);
+}
+
+void SpatializerPoseController::setDesiredMode(HeadTrackingMode mode) {
+    std::lock_guard lock(mMutex);
+    mProcessor->setDesiredMode(mode);
+}
+
+void SpatializerPoseController::setScreenToStagePose(const Pose3f& screenToStage) {
+    std::lock_guard lock(mMutex);
+    mProcessor->setScreenToStagePose(screenToStage);
+}
+
+void SpatializerPoseController::setDisplayOrientation(float physicalToLogicalAngle) {
+    std::lock_guard lock(mMutex);
+    mProcessor->setDisplayOrientation(physicalToLogicalAngle);
+}
+
+void SpatializerPoseController::calculateAsync() {
+    std::lock_guard lock(mMutex);
+    mShouldCalculate = true;
+    mCondVar.notify_all();
+}
+
+void SpatializerPoseController::waitUntilCalculated() {
+    std::unique_lock lock(mMutex);
+    mCondVar.wait(lock, [this] { return mCalculated; });
+}
+
+std::tuple<media::Pose3f, std::optional<media::HeadTrackingMode>>
+SpatializerPoseController::calculate_l() {
+    Pose3f headToStage;
+    HeadTrackingMode mode;
+    std::optional<media::HeadTrackingMode> modeIfChanged;
+
+    mProcessor->calculate(elapsedRealtimeNano());
+    headToStage = mProcessor->getHeadToStagePose();
+    mode = mProcessor->getActualMode();
+    if (!mActualMode.has_value() || mActualMode.value() != mode) {
+        mActualMode = mode;
+        modeIfChanged = mode;
+    }
+    return std::make_tuple(headToStage, modeIfChanged);
+}
+
+void SpatializerPoseController::recenter() {
+    std::lock_guard lock(mMutex);
+    mProcessor->recenter();
+}
+
+void SpatializerPoseController::onPose(int64_t timestamp, int32_t sensor, const Pose3f& pose,
+                                       const std::optional<Twist3f>& twist) {
+    std::lock_guard lock(mMutex);
+    if (sensor == mHeadSensor) {
+        mProcessor->setWorldToHeadPose(timestamp, pose, twist.value_or(Twist3f()));
+    }
+    if (sensor == mScreenSensor) {
+        mProcessor->setWorldToScreenPose(timestamp, pose);
+    }
+}
+
+}  // namespace android
diff --git a/services/audiopolicy/service/SpatializerPoseController.h b/services/audiopolicy/service/SpatializerPoseController.h
new file mode 100644
index 0000000..12898aa
--- /dev/null
+++ b/services/audiopolicy/service/SpatializerPoseController.h
@@ -0,0 +1,140 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#pragma once
+
+#include <chrono>
+#include <condition_variable>
+#include <limits>
+#include <memory>
+#include <mutex>
+#include <thread>
+
+#include <media/HeadTrackingProcessor.h>
+#include <media/SensorPoseProvider.h>
+
+namespace android {
+
+/**
+ * This class encapsulates the logic for pose processing, intended for driving a spatializer effect.
+ * This includes integration with the Sensor sub-system for retrieving sensor data, doing all the
+ * necessary processing, etc.
+ *
+ * Calculations happen on a dedicated thread and published to the client via the Listener interface.
+ * A calculation may be triggered in one of two ways:
+ * - By calling calculateAsync() - calculation will be kicked off in the background.
+ * - By setting a timeout in the ctor, a calculation will be triggered after the timeout elapsed
+ *   from the last calculateAsync() call.
+ *
+ * This class is thread-safe.
+ */
+class SpatializerPoseController : private media::SensorPoseProvider::Listener {
+  public:
+    /**
+     * Listener interface for getting pose and mode updates.
+     * Methods will always be invoked from a designated thread.
+     */
+    class Listener {
+      public:
+        virtual ~Listener() = default;
+
+        virtual void onHeadToStagePose(const media::Pose3f&) = 0;
+        virtual void onActualModeChange(media::HeadTrackingMode) = 0;
+    };
+
+    /**
+     * Ctor.
+     * sensorPeriod determines how often to receive updates from the sensors (input rate).
+     * maxUpdatePeriod determines how often to produce an output when calculateAsync() isn't
+     * invoked.
+     */
+    SpatializerPoseController(Listener* listener, std::chrono::microseconds sensorPeriod,
+                               std::chrono::microseconds maxUpdatePeriod);
+
+    /** Dtor. */
+    ~SpatializerPoseController();
+
+    /**
+     * Set the sensor that is to be used for head-tracking.
+     * nullptr can be used to disable head-tracking.
+     */
+    void setHeadSensor(const ASensor* sensor);
+
+    /**
+     * Set the sensor that is to be used for screen-tracking.
+     * nullptr can be used to disable screen-tracking.
+     */
+    void setScreenSensor(const ASensor* sensor);
+
+    /** Sets the desired head-tracking mode. */
+    void setDesiredMode(media::HeadTrackingMode mode);
+
+    /**
+     * Set the screen-to-stage pose, used in all modes.
+     */
+    void setScreenToStagePose(const media::Pose3f& screenToStage);
+
+    /**
+     * Sets the display orientation.
+     * Orientation is expressed in the angle of rotation from the physical "up" side of the screen
+     * to the logical "up" side of the content displayed the screen. Counterclockwise angles, as
+     * viewed while facing the screen are positive.
+     */
+    void setDisplayOrientation(float physicalToLogicalAngle);
+
+    /**
+     * This causes the current poses for both the head and screen to be considered "center".
+     */
+    void recenter();
+
+    /**
+     * This call triggers the recalculation of the output and the invocation of the relevant
+     * callbacks. This call is async and the callbacks will be triggered shortly after.
+     */
+    void calculateAsync();
+
+    /**
+     * Blocks until calculation and invocation of the respective callbacks has happened at least
+     * once. Do not call from within callbacks.
+     */
+    void waitUntilCalculated();
+
+  private:
+    mutable std::mutex mMutex;
+    Listener* const mListener;
+    const std::chrono::microseconds mSensorPeriod;
+    // Order matters for the following two members to ensure correct destruction.
+    std::unique_ptr<media::HeadTrackingProcessor> mProcessor;
+    std::unique_ptr<media::SensorPoseProvider> mPoseProvider;
+    int32_t mHeadSensor = media::SensorPoseProvider::INVALID_HANDLE;
+    int32_t mScreenSensor = media::SensorPoseProvider::INVALID_HANDLE;
+    std::optional<media::HeadTrackingMode> mActualMode;
+    std::thread mThread;
+    std::condition_variable mCondVar;
+    bool mShouldCalculate = true;
+    bool mShouldExit = false;
+    bool mCalculated = false;
+
+    void onPose(int64_t timestamp, int32_t sensor, const media::Pose3f& pose,
+                const std::optional<media::Twist3f>& twist) override;
+
+    /**
+     * Calculates the new outputs and updates internal state. Must be called with the lock held.
+     * Returns values that should be passed to the respective callbacks.
+     */
+    std::tuple<media::Pose3f, std::optional<media::HeadTrackingMode>> calculate_l();
+};
+
+}  // namespace android
diff --git a/services/audiopolicy/tests/AudioPolicyManagerTestClient.h b/services/audiopolicy/tests/AudioPolicyManagerTestClient.h
index f7b0565..84b40d2 100644
--- a/services/audiopolicy/tests/AudioPolicyManagerTestClient.h
+++ b/services/audiopolicy/tests/AudioPolicyManagerTestClient.h
@@ -37,7 +37,8 @@
 
     status_t openOutput(audio_module_handle_t module,
                         audio_io_handle_t *output,
-                        audio_config_t * /*config*/,
+                        audio_config_t * /*halConfig*/,
+                        audio_config_base_t * /*mixerConfig*/,
                         const sp<DeviceDescriptorBase>& /*device*/,
                         uint32_t * /*latencyMs*/,
                         audio_output_flags_t /*flags*/) override {
diff --git a/services/audiopolicy/tests/AudioPolicyTestClient.h b/services/audiopolicy/tests/AudioPolicyTestClient.h
index 1384864..4e0735b 100644
--- a/services/audiopolicy/tests/AudioPolicyTestClient.h
+++ b/services/audiopolicy/tests/AudioPolicyTestClient.h
@@ -30,7 +30,8 @@
     }
     status_t openOutput(audio_module_handle_t /*module*/,
                         audio_io_handle_t* /*output*/,
-                        audio_config_t* /*config*/,
+                        audio_config_t* /*halConfig*/,
+                        audio_config_base_t* /*mixerConfig*/,
                         const sp<DeviceDescriptorBase>& /*device*/,
                         uint32_t* /*latencyMs*/,
                         audio_output_flags_t /*flags*/) override { return NO_INIT; }
diff --git a/services/camera/libcameraservice/CameraService.cpp b/services/camera/libcameraservice/CameraService.cpp
index d0d3a9d..8ce732f 100644
--- a/services/camera/libcameraservice/CameraService.cpp
+++ b/services/camera/libcameraservice/CameraService.cpp
@@ -2178,7 +2178,6 @@
     newDeviceState |= vendorBits;
 
     ALOGV("%s: New device state 0x%" PRIx64, __FUNCTION__, newDeviceState);
-    Mutex::Autolock l(mServiceLock);
     mCameraProviderManager->notifyDeviceStateChange(newDeviceState);
 
     return Status::ok();
diff --git a/services/camera/libcameraservice/common/CameraProviderManager.cpp b/services/camera/libcameraservice/common/CameraProviderManager.cpp
index 4f2b878..556ddda 100644
--- a/services/camera/libcameraservice/common/CameraProviderManager.cpp
+++ b/services/camera/libcameraservice/common/CameraProviderManager.cpp
@@ -359,7 +359,13 @@
     for (auto& provider : mProviders) {
         ALOGV("%s: Notifying %s for new state 0x%" PRIx64,
                 __FUNCTION__, provider->mProviderName.c_str(), newState);
+        // b/199240726 Camera providers can for example try to add/remove
+        // camera devices as part of the state change notification. Holding
+        // 'mInterfaceMutex' while calling 'notifyDeviceStateChange' can
+        // result in a recursive deadlock.
+        mInterfaceMutex.unlock();
         status_t singleRes = provider->notifyDeviceStateChange(mDeviceState);
+        mInterfaceMutex.lock();
         if (singleRes != OK) {
             ALOGE("%s: Unable to notify provider %s about device state change",
                     __FUNCTION__,
@@ -1185,10 +1191,12 @@
 }
 
 bool CameraProviderManager::isHiddenPhysicalCamera(const std::string& cameraId) const {
+    std::lock_guard<std::mutex> lock(mInterfaceMutex);
     return isHiddenPhysicalCameraInternal(cameraId).first;
 }
 
 status_t CameraProviderManager::filterSmallJpegSizes(const std::string& cameraId) {
+    std::lock_guard<std::mutex> lock(mInterfaceMutex);
     for (auto& provider : mProviders) {
         for (auto& deviceInfo : provider->mDevices) {
             if (deviceInfo->mId == cameraId) {
diff --git a/services/camera/libcameraservice/common/CameraProviderManager.h b/services/camera/libcameraservice/common/CameraProviderManager.h
index 1bdbb44..d8c1f59 100644
--- a/services/camera/libcameraservice/common/CameraProviderManager.h
+++ b/services/camera/libcameraservice/common/CameraProviderManager.h
@@ -284,12 +284,6 @@
             sp<hardware::camera::device::V3_2::ICameraDeviceSession> *session);
 
     /**
-     * Save the ICameraProvider while it is being used by a camera or torch client
-     */
-    void saveRef(DeviceMode usageType, const std::string &cameraId,
-            sp<hardware::camera::provider::V2_4::ICameraProvider> provider);
-
-    /**
      * Notify that the camera or torch is no longer being used by a camera client
      */
     void removeRef(DeviceMode usageType, const std::string &cameraId);
@@ -434,6 +428,10 @@
 
         /**
          * Notify provider about top-level device physical state changes
+         *
+         * Note that 'mInterfaceMutex' should not be held when calling this method.
+         * It is possible for camera providers to add/remove devices and try to
+         * acquire it.
          */
         status_t notifyDeviceStateChange(
                 hardware::hidl_bitfield<hardware::camera::provider::V2_5::DeviceState>
@@ -662,6 +660,12 @@
                 sp<hardware::camera::provider::V2_6::ICameraProvider> &interface2_6);
     };
 
+    /**
+     * Save the ICameraProvider while it is being used by a camera or torch client
+     */
+    void saveRef(DeviceMode usageType, const std::string &cameraId,
+            sp<hardware::camera::provider::V2_4::ICameraProvider> provider);
+
     // Utility to find a DeviceInfo by ID; pointer is only valid while mInterfaceMutex is held
     // and the calling code doesn't mutate the list of providers or their lists of devices.
     // Finds the first device of the given ID that falls within the requested version range
diff --git a/services/mediametrics/Android.bp b/services/mediametrics/Android.bp
index 5989181..0351d2d 100644
--- a/services/mediametrics/Android.bp
+++ b/services/mediametrics/Android.bp
@@ -148,7 +148,8 @@
         "statsd_mediaparser.cpp",
         "statsd_nuplayer.cpp",
         "statsd_recorder.cpp",
-        "StringUtils.cpp"
+        "StringUtils.cpp",
+        "ValidateId.cpp",
     ],
 
     proto: {
diff --git a/services/mediametrics/AudioAnalytics.cpp b/services/mediametrics/AudioAnalytics.cpp
index 45c9f56..270fe2f 100644
--- a/services/mediametrics/AudioAnalytics.cpp
+++ b/services/mediametrics/AudioAnalytics.cpp
@@ -29,6 +29,7 @@
 #include "AudioTypes.h"           // string to int conversions
 #include "MediaMetricsService.h"  // package info
 #include "StringUtils.h"
+#include "ValidateId.h"
 
 #define PROP_AUDIO_ANALYTICS_CLOUD_ENABLED "persist.audio.analytics.cloud.enabled"
 
@@ -563,7 +564,7 @@
         const auto flagsForStats = types::lookup<types::INPUT_FLAG, short_enum_type_t>(flags);
         const auto sourceForStats = types::lookup<types::SOURCE_TYPE, short_enum_type_t>(source);
         // Android S
-        const auto logSessionIdForStats = stringutils::sanitizeLogSessionId(logSessionId);
+        const auto logSessionIdForStats = ValidateId::get()->validateId(logSessionId);
 
         LOG(LOG_LEVEL) << "key:" << key
               << " id:" << id
@@ -718,7 +719,7 @@
                  types::lookup<types::TRACK_TRAITS, short_enum_type_t>(traits);
         const auto usageForStats = types::lookup<types::USAGE, short_enum_type_t>(usage);
         // Android S
-        const auto logSessionIdForStats = stringutils::sanitizeLogSessionId(logSessionId);
+        const auto logSessionIdForStats = ValidateId::get()->validateId(logSessionId);
 
         LOG(LOG_LEVEL) << "key:" << key
               << " id:" << id
diff --git a/services/mediametrics/LruSet.h b/services/mediametrics/LruSet.h
new file mode 100644
index 0000000..1f0ab60
--- /dev/null
+++ b/services/mediametrics/LruSet.h
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include <list>
+#include <sstream>
+#include <unordered_map>
+
+namespace android::mediametrics {
+
+/**
+ * LruSet keeps a set of the last "Size" elements added or accessed.
+ *
+ * (Lru stands for least-recently-used eviction policy).
+ *
+ * Runs in O(1) time for add, remove, and check.  Internally implemented
+ * with an unordered_map and a list.  In order to remove elements,
+ * a list iterator is stored in the unordered_map
+ * (noting that std::list::erase() contractually
+ * does not affect iterators other than the one erased).
+ */
+
+template <typename T>
+class LruSet {
+    const size_t mMaxSize;
+    std::list<T> mAccessOrder;                 // front is the most recent, back is the oldest.
+    // item T with its access order iterator.
+    std::unordered_map<T, typename std::list<T>::iterator> mMap;
+
+public:
+    /**
+     * Constructs a LruSet which checks whether the element was
+     * accessed or added recently.
+     *
+     * The parameter maxSize is used to cap growth of LruSet;
+     * eviction is based on least recently used LRU.
+     * If maxSize is zero, the LruSet contains no elements
+     * and check() always returns false.
+     *
+     * \param maxSize the maximum number of elements that are tracked.
+     */
+    explicit LruSet(size_t maxSize) : mMaxSize(maxSize) {}
+
+    /**
+     * Returns the number of entries in the LruSet.
+     *
+     * This is a number between 0 and maxSize.
+     */
+    size_t size() const {
+        return mMap.size();
+    }
+
+    /** Clears the container contents. */
+    void clear() {
+        mMap.clear();
+        mAccessOrder.clear();
+    }
+
+    /** Returns a string dump of the last n entries. */
+    std::string dump(size_t n) const {
+        std::stringstream ss;
+        auto it = mAccessOrder.cbegin();
+        for (size_t i = 0; i < n && it != mAccessOrder.cend(); ++i) {
+            ss << *it++ << "\n";
+        }
+        return ss.str();
+    }
+
+    /** Adds a new item to the set. */
+    void add(const T& t) {
+        if (mMaxSize == 0) return;
+        auto it = mMap.find(t);
+        if (it != mMap.end()) { // already exists.
+            mAccessOrder.erase(it->second);  // move-to-front on the chronologically ordered list.
+        } else if (mAccessOrder.size() >= mMaxSize) {
+            const T last = mAccessOrder.back();
+            mAccessOrder.pop_back();
+            mMap.erase(last);
+        }
+        mAccessOrder.push_front(t);
+        mMap[t] = mAccessOrder.begin();
+    }
+
+    /**
+     * Removes an item from the set.
+     *
+     * \param t item to be removed.
+     * \return false if the item doesn't exist.
+     */
+    bool remove(const T& t) {
+        auto it = mMap.find(t);
+        if (it == mMap.end()) return false;
+        mAccessOrder.erase(it->second);
+        mMap.erase(it);
+        return true;
+    }
+
+    /** Returns true if t is present (and moves the access order of t to the front). */
+    bool check(const T& t) { // not const, as it adjusts the least-recently-used order.
+        auto it = mMap.find(t);
+        if (it == mMap.end()) return false;
+        mAccessOrder.erase(it->second);
+        mAccessOrder.push_front(it->first);
+        it->second = mAccessOrder.begin();
+        return true;
+    }
+};
+
+} // namespace android::mediametrics
diff --git a/services/mediametrics/MediaMetricsService.cpp b/services/mediametrics/MediaMetricsService.cpp
index 1d64878..35e0ae4 100644
--- a/services/mediametrics/MediaMetricsService.cpp
+++ b/services/mediametrics/MediaMetricsService.cpp
@@ -19,6 +19,7 @@
 #include <utils/Log.h>
 
 #include "MediaMetricsService.h"
+#include "ValidateId.h"
 #include "iface_statsd.h"
 
 #include <pwd.h> //getpwuid
@@ -204,6 +205,15 @@
     // now attach either the item or its dup to a const shared pointer
     std::shared_ptr<const mediametrics::Item> sitem(release ? item : item->dup());
 
+    // register log session ids with singleton.
+    if (startsWith(item->getKey(), "metrics.manager")) {
+        std::string logSessionId;
+        if (item->get("logSessionId", &logSessionId)
+                && mediametrics::stringutils::isLogSessionId(logSessionId.c_str())) {
+            mediametrics::ValidateId::get()->registerId(logSessionId);
+        }
+    }
+
     (void)mAudioAnalytics.submit(sitem, isTrusted);
 
     (void)dump2Statsd(sitem, mStatsdLog);  // failure should be logged in function.
@@ -309,6 +319,9 @@
                 result << "-- some lines may be truncated --\n";
             }
 
+            result << "LogSessionId:\n"
+                   << mediametrics::ValidateId::get()->dump();
+
             // Dump the statsd atoms we sent out.
             result << "Statsd atoms:\n"
                    << mStatsdLog->dumpToString("  " /* prefix */,
diff --git a/services/mediametrics/ValidateId.cpp b/services/mediametrics/ValidateId.cpp
new file mode 100644
index 0000000..0cc8593
--- /dev/null
+++ b/services/mediametrics/ValidateId.cpp
@@ -0,0 +1,65 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+//#define LOG_NDEBUG 0
+#define LOG_TAG "MediaMetricsService"  // not ValidateId
+#include <utils/Log.h>
+
+#include "ValidateId.h"
+
+namespace android::mediametrics {
+
+std::string ValidateId::dump() const
+{
+    std::stringstream ss;
+    ss << "Entries:" << mIdSet.size() << "  InvalidIds:" << mInvalidIds << "\n";
+    ss << mIdSet.dump(10);
+    return ss.str();
+}
+
+void ValidateId::registerId(const std::string& id)
+{
+    if (id.empty()) return;
+    if (!mediametrics::stringutils::isLogSessionId(id.c_str())) {
+        ALOGW("%s: rejecting malformed id %s", __func__, id.c_str());
+        return;
+    }
+    ALOGV("%s: registering %s", __func__, id.c_str());
+    mIdSet.add(id);
+}
+
+const std::string& ValidateId::validateId(const std::string& id)
+{
+    static const std::string empty{};
+    if (id.empty()) return empty;
+
+    // reject because the id is malformed
+    if (!mediametrics::stringutils::isLogSessionId(id.c_str())) {
+        ALOGW("%s: rejecting malformed id %s", __func__, id.c_str());
+        ++mInvalidIds;
+        return empty;
+    }
+
+    // reject because the id is unregistered
+    if (!mIdSet.check(id)) {
+        ALOGW("%s: rejecting unregistered id %s", __func__, id.c_str());
+        ++mInvalidIds;
+        return empty;
+    }
+    return id;
+}
+
+} // namespace android::mediametrics
diff --git a/services/mediametrics/ValidateId.h b/services/mediametrics/ValidateId.h
new file mode 100644
index 0000000..166b39a
--- /dev/null
+++ b/services/mediametrics/ValidateId.h
@@ -0,0 +1,80 @@
+/*
+ * Copyright (C) 2021 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include "LruSet.h"
+#include "StringUtils.h"
+#include "Wrap.h"
+
+namespace android::mediametrics {
+
+/*
+ * ValidateId is used to check whether the log session id is properly formed
+ * and has been registered (i.e. from the Java MediaMetricsManagerService).
+ *
+ * The default memory window to track registered ids is set to SINGLETON_LRU_SET_SIZE.
+ *
+ * This class is not thread-safe, but the singleton returned by get() uses LockWrap<>
+ * to ensure thread-safety.
+ */
+class ValidateId {
+    mediametrics::LruSet<std::string> mIdSet;
+    size_t mInvalidIds = 0;  // count invalid ids encountered.
+public:
+    /** Creates a ValidateId object with size memory window. */
+    explicit ValidateId(size_t size) : mIdSet{size} {}
+
+    /** Returns a string dump of recent contents and stats. */
+    std::string dump() const;
+
+    /**
+     * Registers the id string.
+     *
+     * If id string is malformed (not 16 Base64Url chars), it is ignored.
+     * Once registered, calling validateId() will return id (instead of the empty string).
+     * ValidateId may "forget" the id after not encountering it within the past N ids,
+     * where N is the size set in the constructor.
+     *
+     * param id string (from MediaMetricsManagerService).
+     */
+    void registerId(const std::string& id);
+
+    /**
+     * Returns the empty string if id string is malformed (not 16 Base64Url chars)
+     * or if id string has not been seen (in the recent size ids);
+     * otherwise it returns the same id parameter.
+     *
+     * \param id string (to be sent to statsd).
+     */
+    const std::string& validateId(const std::string& id);
+
+    /** Singleton set size */
+    static inline constexpr size_t SINGLETON_LRU_SET_SIZE = 2000;
+
+    using LockedValidateId = mediametrics::LockWrap<ValidateId>;
+    /**
+     * Returns a singleton locked ValidateId object that is thread-safe using LockWrap<>.
+     *
+     * The Singleton ValidateId object is created with size LRU_SET_SIZE (during first call).
+     */
+    static inline LockedValidateId& get() {
+        static LockedValidateId privateSet{SINGLETON_LRU_SET_SIZE};
+        return privateSet;
+    }
+};
+
+} // namespace android::mediametrics
diff --git a/services/mediametrics/statsd_audiorecord.cpp b/services/mediametrics/statsd_audiorecord.cpp
index 41efcaa..c53b6f3 100644
--- a/services/mediametrics/statsd_audiorecord.cpp
+++ b/services/mediametrics/statsd_audiorecord.cpp
@@ -32,7 +32,7 @@
 #include <statslog.h>
 
 #include "MediaMetricsService.h"
-#include "StringUtils.h"
+#include "ValidateId.h"
 #include "frameworks/proto_logging/stats/message/mediametrics_message.pb.h"
 #include "iface_statsd.h"
 
@@ -143,8 +143,7 @@
     // log_session_id (string)
     std::string logSessionId;
     (void)item->getString("android.media.audiorecord.logSessionId", &logSessionId);
-    const auto log_session_id =
-            mediametrics::stringutils::sanitizeLogSessionId(logSessionId);
+    const auto log_session_id = mediametrics::ValidateId::get()->validateId(logSessionId);
 
     android::util::BytesField bf_serialized( serialized.c_str(), serialized.size());
     int result = android::util::stats_write(android::util::MEDIAMETRICS_AUDIORECORD_REPORTED,
diff --git a/services/mediametrics/statsd_audiotrack.cpp b/services/mediametrics/statsd_audiotrack.cpp
index 59627ae..707effd 100644
--- a/services/mediametrics/statsd_audiotrack.cpp
+++ b/services/mediametrics/statsd_audiotrack.cpp
@@ -32,7 +32,7 @@
 #include <statslog.h>
 
 #include "MediaMetricsService.h"
-#include "StringUtils.h"
+#include "ValidateId.h"
 #include "frameworks/proto_logging/stats/message/mediametrics_message.pb.h"
 #include "iface_statsd.h"
 
@@ -137,8 +137,7 @@
     // log_session_id (string)
     std::string logSessionId;
     (void)item->getString("android.media.audiotrack.logSessionId", &logSessionId);
-    const auto log_session_id =
-            mediametrics::stringutils::sanitizeLogSessionId(logSessionId);
+    const auto log_session_id = mediametrics::ValidateId::get()->validateId(logSessionId);
 
     android::util::BytesField bf_serialized( serialized.c_str(), serialized.size());
     int result = android::util::stats_write(android::util::MEDIAMETRICS_AUDIOTRACK_REPORTED,
diff --git a/services/mediametrics/statsd_codec.cpp b/services/mediametrics/statsd_codec.cpp
index 46cbdc8..8581437 100644
--- a/services/mediametrics/statsd_codec.cpp
+++ b/services/mediametrics/statsd_codec.cpp
@@ -34,7 +34,7 @@
 
 #include "cleaner.h"
 #include "MediaMetricsService.h"
-#include "StringUtils.h"
+#include "ValidateId.h"
 #include "frameworks/proto_logging/stats/message/mediametrics_message.pb.h"
 #include "iface_statsd.h"
 
@@ -228,7 +228,7 @@
 
     std::string sessionId;
     if (item->getString("android.media.mediacodec.log-session-id", &sessionId)) {
-        sessionId = mediametrics::stringutils::sanitizeLogSessionId(sessionId);
+        sessionId = mediametrics::ValidateId::get()->validateId(sessionId);
         metrics_proto.set_log_session_id(sessionId);
     }
     AStatsEvent_writeString(event, codec.c_str());
diff --git a/services/mediametrics/statsd_extractor.cpp b/services/mediametrics/statsd_extractor.cpp
index bcf2e0a..a8bfeaa 100644
--- a/services/mediametrics/statsd_extractor.cpp
+++ b/services/mediametrics/statsd_extractor.cpp
@@ -32,7 +32,7 @@
 #include <statslog.h>
 
 #include "MediaMetricsService.h"
-#include "StringUtils.h"
+#include "ValidateId.h"
 #include "frameworks/proto_logging/stats/message/mediametrics_message.pb.h"
 #include "iface_statsd.h"
 
@@ -86,7 +86,7 @@
 
     std::string log_session_id;
     if (item->getString("android.media.mediaextractor.logSessionId", &log_session_id)) {
-        log_session_id = mediametrics::stringutils::sanitizeLogSessionId(log_session_id);
+        log_session_id = mediametrics::ValidateId::get()->validateId(log_session_id);
         metrics_proto.set_log_session_id(log_session_id);
     }
 
diff --git a/services/mediametrics/statsd_mediaparser.cpp b/services/mediametrics/statsd_mediaparser.cpp
index 921b320..67ca874b 100644
--- a/services/mediametrics/statsd_mediaparser.cpp
+++ b/services/mediametrics/statsd_mediaparser.cpp
@@ -31,7 +31,7 @@
 #include <statslog.h>
 
 #include "MediaMetricsService.h"
-#include "StringUtils.h"
+#include "ValidateId.h"
 #include "frameworks/proto_logging/stats/enums/stats/mediametrics/mediametrics.pb.h"
 #include "iface_statsd.h"
 
@@ -81,7 +81,7 @@
 
     std::string logSessionId;
     item->getString("android.media.mediaparser.logSessionId", &logSessionId);
-    logSessionId = mediametrics::stringutils::sanitizeLogSessionId(logSessionId);
+    logSessionId = mediametrics::ValidateId::get()->validateId(logSessionId);
 
     int result = android::util::stats_write(android::util::MEDIAMETRICS_MEDIAPARSER_REPORTED,
                                timestamp_nanos,
diff --git a/services/mediametrics/statsd_recorder.cpp b/services/mediametrics/statsd_recorder.cpp
index b29ad73..5f54a68 100644
--- a/services/mediametrics/statsd_recorder.cpp
+++ b/services/mediametrics/statsd_recorder.cpp
@@ -32,7 +32,7 @@
 #include <statslog.h>
 
 #include "MediaMetricsService.h"
-#include "StringUtils.h"
+#include "ValidateId.h"
 #include "frameworks/proto_logging/stats/message/mediametrics_message.pb.h"
 #include "iface_statsd.h"
 
@@ -59,7 +59,7 @@
     // string kRecorderLogSessionId = "android.media.mediarecorder.log-session-id";
     std::string log_session_id;
     if (item->getString("android.media.mediarecorder.log-session-id", &log_session_id)) {
-        log_session_id = mediametrics::stringutils::sanitizeLogSessionId(log_session_id);
+        log_session_id = mediametrics::ValidateId::get()->validateId(log_session_id);
         metrics_proto.set_log_session_id(log_session_id);
     }
     // string kRecorderAudioMime = "android.media.mediarecorder.audio.mime";
diff --git a/services/mediametrics/tests/mediametrics_tests.cpp b/services/mediametrics/tests/mediametrics_tests.cpp
index 2336d6f..69ec947 100644
--- a/services/mediametrics/tests/mediametrics_tests.cpp
+++ b/services/mediametrics/tests/mediametrics_tests.cpp
@@ -28,6 +28,7 @@
 
 #include "AudioTypes.h"
 #include "StringUtils.h"
+#include "ValidateId.h"
 
 using namespace android;
 
@@ -1127,3 +1128,100 @@
    validId2[3] = '!';
    ASSERT_EQ("", mediametrics::stringutils::sanitizeLogSessionId(validId2));
 }
+
+TEST(mediametrics_tests, LruSet) {
+    constexpr size_t LRU_SET_SIZE = 2;
+    mediametrics::LruSet<std::string> lruSet(LRU_SET_SIZE);
+
+    // test adding a couple strings.
+    lruSet.add("abc");
+    ASSERT_EQ(1u, lruSet.size());
+    ASSERT_TRUE(lruSet.check("abc"));
+    lruSet.add("def");
+    ASSERT_EQ(2u, lruSet.size());
+
+    // now adding the third string causes eviction of the oldest.
+    lruSet.add("ghi");
+    ASSERT_FALSE(lruSet.check("abc"));
+    ASSERT_TRUE(lruSet.check("ghi"));
+    ASSERT_TRUE(lruSet.check("def"));  // "def" is most recent.
+    ASSERT_EQ(2u, lruSet.size());      // "abc" is correctly discarded.
+
+    // adding another string will evict the oldest.
+    lruSet.add("foo");
+    ASSERT_FALSE(lruSet.check("ghi")); // note: "ghi" discarded when "foo" added.
+    ASSERT_TRUE(lruSet.check("foo"));
+    ASSERT_TRUE(lruSet.check("def"));
+
+    // manual removing of a string works, too.
+    ASSERT_TRUE(lruSet.remove("def"));
+    ASSERT_FALSE(lruSet.check("def")); // we manually removed "def".
+    ASSERT_TRUE(lruSet.check("foo"));  // "foo" is still there.
+    ASSERT_EQ(1u, lruSet.size());
+
+    // you can't remove a string that has not been added.
+    ASSERT_FALSE(lruSet.remove("bar")); // Note: "bar" doesn't exist, so remove returns false.
+    ASSERT_EQ(1u, lruSet.size());
+
+    lruSet.add("foo");   // adding "foo" (which already exists) doesn't change size.
+    ASSERT_EQ(1u, lruSet.size());
+    lruSet.add("bar");   // add "bar"
+    ASSERT_EQ(2u, lruSet.size());
+    lruSet.add("glorp"); // add "glorp" evicts "foo".
+    ASSERT_EQ(2u, lruSet.size());
+    ASSERT_TRUE(lruSet.check("bar"));
+    ASSERT_TRUE(lruSet.check("glorp"));
+    ASSERT_FALSE(lruSet.check("foo"));
+}
+
+TEST(mediametrics_tests, LruSet0) {
+    constexpr size_t LRU_SET_SIZE = 0;
+    mediametrics::LruSet<std::string> lruSet(LRU_SET_SIZE);
+
+    lruSet.add("a");
+    ASSERT_EQ(0u, lruSet.size());
+    ASSERT_FALSE(lruSet.check("a"));
+    ASSERT_FALSE(lruSet.remove("a")); // never added.
+    ASSERT_EQ(0u, lruSet.size());
+}
+
+// Returns a 16 Base64Url string representing the decimal representation of value
+// (with leading 0s) e.g. 0000000000000000, 0000000000000001, 0000000000000002, ...
+static std::string generateId(size_t value)
+{
+    char id[16 + 1]; // to be filled with 16 Base64Url chars (and zero termination)
+    char *sptr = id + 16; // start at the end.
+    *sptr-- = 0; // zero terminate.
+    // output the digits from least significant to most significant.
+    while (value) {
+        *sptr-- = value % 10;
+        value /= 10;
+    }
+    // add leading 0's
+    while (sptr > id) {
+        *sptr-- = '0';
+    }
+    return std::string(id);
+}
+
+TEST(mediametrics_tests, ValidateId) {
+    constexpr size_t LRU_SET_SIZE = 3;
+    constexpr size_t IDS = 10;
+    static_assert(IDS > LRU_SET_SIZE);  // IDS must be greater than LRU_SET_SIZE.
+    mediametrics::ValidateId validateId(LRU_SET_SIZE);
+
+
+    // register IDs as integer strings counting from 0.
+    for (size_t i = 0; i < IDS; ++i) {
+        validateId.registerId(generateId(i));
+    }
+
+    // only the last LRU_SET_SIZE exist.
+    for (size_t i = 0; i < IDS - LRU_SET_SIZE; ++i) {
+        ASSERT_EQ("", validateId.validateId(generateId(i)));
+    }
+    for (size_t i = IDS - LRU_SET_SIZE; i < IDS; ++i) {
+        const std::string id = generateId(i);
+        ASSERT_EQ(id, validateId.validateId(id));
+    }
+}
diff --git a/services/oboeservice/AAudioServiceEndpoint.cpp b/services/oboeservice/AAudioServiceEndpoint.cpp
index 13dd3d3..390cd5c 100644
--- a/services/oboeservice/AAudioServiceEndpoint.cpp
+++ b/services/oboeservice/AAudioServiceEndpoint.cpp
@@ -59,6 +59,7 @@
     result << "    Device Id:            " << getDeviceId() << "\n";
     result << "    Sample Rate:          " << getSampleRate() << "\n";
     result << "    Channel Count:        " << getSamplesPerFrame() << "\n";
+    result << "    Channel Mask:         0x" << std::hex << getChannelMask() << std::dec << "\n";
     result << "    Format:               " << getFormat() << "\n";
     result << "    Frames Per Burst:     " << mFramesPerBurst << "\n";
     result << "    Usage:                " << getUsage() << "\n";
@@ -164,6 +165,10 @@
         configuration.getSamplesPerFrame() != getSamplesPerFrame()) {
         return false;
     }
+    if (configuration.getChannelMask() != AAUDIO_UNSPECIFIED &&
+        configuration.getChannelMask() != getChannelMask()) {
+        return false;
+    }
     return true;
 }
 
@@ -188,7 +193,9 @@
     if (direction == AAUDIO_DIRECTION_OUTPUT) {
         flags = static_cast<audio_flags_mask_t>(AUDIO_FLAG_LOW_LATENCY
                 | AAudioConvert_allowCapturePolicyToAudioFlagsMask(
-                        params->getAllowedCapturePolicy()));
+                        params->getAllowedCapturePolicy(),
+                        params->getSpatializationBehavior(),
+                        params->isContentSpatialized()));
     } else {
         flags = static_cast<audio_flags_mask_t>(AUDIO_FLAG_LOW_LATENCY
                 | AAudioConvert_privacySensitiveToAudioFlagsMask(params->isPrivacySensitive()));
diff --git a/services/oboeservice/AAudioServiceEndpointMMAP.cpp b/services/oboeservice/AAudioServiceEndpointMMAP.cpp
index a08098c..35a0890 100644
--- a/services/oboeservice/AAudioServiceEndpointMMAP.cpp
+++ b/services/oboeservice/AAudioServiceEndpointMMAP.cpp
@@ -126,20 +126,15 @@
     }
     config.sample_rate = aaudioSampleRate;
 
-    int32_t aaudioSamplesPerFrame = getSamplesPerFrame();
-
     const aaudio_direction_t direction = getDirection();
 
+    config.channel_mask = AAudio_getChannelMaskForOpen(
+            getChannelMask(), getSamplesPerFrame(), direction == AAUDIO_DIRECTION_INPUT);
+
     if (direction == AAUDIO_DIRECTION_OUTPUT) {
-        config.channel_mask = (aaudioSamplesPerFrame == AAUDIO_UNSPECIFIED)
-                              ? AUDIO_CHANNEL_OUT_STEREO
-                              : audio_channel_out_mask_from_count(aaudioSamplesPerFrame);
         mHardwareTimeOffsetNanos = OUTPUT_ESTIMATED_HARDWARE_OFFSET_NANOS; // frames at DAC later
 
     } else if (direction == AAUDIO_DIRECTION_INPUT) {
-        config.channel_mask =  (aaudioSamplesPerFrame == AAUDIO_UNSPECIFIED)
-                               ? AUDIO_CHANNEL_IN_STEREO
-                               : audio_channel_in_mask_from_count(aaudioSamplesPerFrame);
         mHardwareTimeOffsetNanos = INPUT_ESTIMATED_HARDWARE_OFFSET_NANOS; // frames at ADC earlier
 
     } else {
@@ -225,9 +220,9 @@
     }
 
     // Get information about the stream and pass it back to the caller.
-    setSamplesPerFrame((direction == AAUDIO_DIRECTION_OUTPUT)
-                       ? audio_channel_count_from_out_mask(config.channel_mask)
-                       : audio_channel_count_from_in_mask(config.channel_mask));
+    setChannelMask(AAudioConvert_androidToAAudioChannelMask(
+            config.channel_mask, getDirection() == AAUDIO_DIRECTION_INPUT,
+            AAudio_isChannelIndexMask(config.channel_mask)));
 
     // AAudio creates a copy of this FD and retains ownership of the copy.
     // Assume that AudioFlinger will close the original shared_memory_fd.
@@ -247,9 +242,9 @@
     setFormat(config.format);
     setSampleRate(config.sample_rate);
 
-    ALOGD("%s() actual rate = %d, channels = %d"
-          ", deviceId = %d, capacity = %d\n",
-          __func__, getSampleRate(), getSamplesPerFrame(), deviceId, getBufferCapacity());
+    ALOGD("%s() actual rate = %d, channels = %d channelMask = %#x, deviceId = %d, capacity = %d\n",
+          __func__, getSampleRate(), getSamplesPerFrame(), getChannelMask(),
+          deviceId, getBufferCapacity());
 
     ALOGD("%s() format = 0x%08x, frame size = %d, burst size = %d",
           __func__, getFormat(), calculateBytesPerFrame(), mFramesPerBurst);
diff --git a/services/oboeservice/AAudioServiceEndpointShared.cpp b/services/oboeservice/AAudioServiceEndpointShared.cpp
index 5fbcadb..5af0a91 100644
--- a/services/oboeservice/AAudioServiceEndpointShared.cpp
+++ b/services/oboeservice/AAudioServiceEndpointShared.cpp
@@ -78,7 +78,7 @@
     result = mStreamInternal->open(builder);
 
     setSampleRate(mStreamInternal->getSampleRate());
-    setSamplesPerFrame(mStreamInternal->getSamplesPerFrame());
+    setChannelMask(mStreamInternal->getChannelMask());
     setDeviceId(mStreamInternal->getDeviceId());
     setSessionId(mStreamInternal->getSessionId());
     setFormat(AUDIO_FORMAT_PCM_FLOAT); // force for mixer
diff --git a/services/oboeservice/AAudioServiceStreamBase.cpp b/services/oboeservice/AAudioServiceStreamBase.cpp
index 34ddd4d..4ffc127 100644
--- a/services/oboeservice/AAudioServiceStreamBase.cpp
+++ b/services/oboeservice/AAudioServiceStreamBase.cpp
@@ -73,7 +73,8 @@
 }
 
 std::string AAudioServiceStreamBase::dumpHeader() {
-    return std::string("    T   Handle   UId   Port Run State Format Burst Chan Capacity");
+    return std::string(
+            "    T   Handle   UId   Port Run State Format Burst Chan Mask     Capacity");
 }
 
 std::string AAudioServiceStreamBase::dump() const {
@@ -88,6 +89,7 @@
     result << std::setw(7) << getFormat();
     result << std::setw(6) << mFramesPerBurst;
     result << std::setw(5) << getSamplesPerFrame();
+    result << std::setw(8) << std::hex << getChannelMask() << std::dec;
     result << std::setw(9) << getBufferCapacity();
 
     return result.str();
diff --git a/services/oboeservice/AAudioServiceStreamShared.cpp b/services/oboeservice/AAudioServiceStreamShared.cpp
index c665cda..ad06d97 100644
--- a/services/oboeservice/AAudioServiceStreamShared.cpp
+++ b/services/oboeservice/AAudioServiceStreamShared.cpp
@@ -164,11 +164,11 @@
         goto error;
     }
 
-    setSamplesPerFrame(configurationInput.getSamplesPerFrame());
-    if (getSamplesPerFrame() == AAUDIO_UNSPECIFIED) {
-        setSamplesPerFrame(endpoint->getSamplesPerFrame());
+    setChannelMask(configurationInput.getChannelMask());
+    if (getChannelMask() == AAUDIO_UNSPECIFIED) {
+        setChannelMask(endpoint->getChannelMask());
     } else if (getSamplesPerFrame() != endpoint->getSamplesPerFrame()) {
-        ALOGD("%s() mSamplesPerFrame = %d, need %d",
+        ALOGD("%s() mSamplesPerFrame = %#x, need %#x",
               __func__, getSamplesPerFrame(), endpoint->getSamplesPerFrame());
         result = AAUDIO_ERROR_OUT_OF_RANGE;
         goto error;
diff --git a/services/oboeservice/fuzzer/README.md b/services/oboeservice/fuzzer/README.md
index 00b85df..ae7af3eb 100644
--- a/services/oboeservice/fuzzer/README.md
+++ b/services/oboeservice/fuzzer/README.md
@@ -15,7 +15,7 @@
 4. InService
 5. DeviceId
 6. SampleRate
-7. SamplesPerFrame
+7. ChannelMask
 8. Direction
 9. SharingMode
 10. Usage
@@ -31,7 +31,7 @@
 | `InService`   | `bool` | Value obtained from FuzzedDataProvider |
 | `DeviceId`   | `INT32_MIN` to `INT32_MAX` | Value obtained from FuzzedDataProvider |
 | `SampleRate`   | `INT32_MIN` to `INT32_MAX` | Value obtained from FuzzedDataProvider |
-| `SamplesPerFrame` | `INT32_MIN` to `INT32_MAX` | Value obtained from FuzzedDataProvider |
+| `ChannelMask` | `AAUDIO_UNSPECIFIED`, `AAUDIO_CHANNEL_INDEX_MASK_1`, `AAUDIO_CHANNEL_INDEX_MASK_2`, `AAUDIO_CHANNEL_INDEX_MASK_3`, `AAUDIO_CHANNEL_INDEX_MASK_4`, `AAUDIO_CHANNEL_INDEX_MASK_5`, `AAUDIO_CHANNEL_INDEX_MASK_6`, `AAUDIO_CHANNEL_INDEX_MASK_7`, `AAUDIO_CHANNEL_INDEX_MASK_8`, `AAUDIO_CHANNEL_INDEX_MASK_9`, `AAUDIO_CHANNEL_INDEX_MASK_10`, `AAUDIO_CHANNEL_INDEX_MASK_11`, `AAUDIO_CHANNEL_INDEX_MASK_12`, `AAUDIO_CHANNEL_INDEX_MASK_13`, `AAUDIO_CHANNEL_INDEX_MASK_14`, `AAUDIO_CHANNEL_INDEX_MASK_15`, `AAUDIO_CHANNEL_INDEX_MASK_16`, `AAUDIO_CHANNEL_INDEX_MASK_17`, `AAUDIO_CHANNEL_INDEX_MASK_18`, `AAUDIO_CHANNEL_INDEX_MASK_19`, `AAUDIO_CHANNEL_INDEX_MASK_20`, `AAUDIO_CHANNEL_INDEX_MASK_21`, `AAUDIO_CHANNEL_INDEX_MASK_22`, `AAUDIO_CHANNEL_INDEX_MASK_23`, `AAUDIO_CHANNEL_INDEX_MASK_24`, `AAUDIO_CHANNEL_MONO`, `AAUDIO_CHANNEL_STEREO`, `AAUDIO_CHANNEL_FRONT_BACK`, `AAUDIO_CHANNEL_2POINT0POINT2`, `AAUDIO_CHANNEL_2POINT1POINT2`, `AAUDIO_CHANNEL_3POINT0POINT2`, `AAUDIO_CHANNEL_3POINT1POINT2`, `AAUDIO_CHANNEL_5POINT1`, `AAUDIO_CHANNEL_MONO`, `AAUDIO_CHANNEL_STEREO`, `AAUDIO_CHANNEL_2POINT1`, `AAUDIO_CHANNEL_TRI`, `AAUDIO_CHANNEL_TRI_BACK`, `AAUDIO_CHANNEL_3POINT1`, `AAUDIO_CHANNEL_2POINT0POINT2`, `AAUDIO_CHANNEL_2POINT1POINT2`, `AAUDIO_CHANNEL_3POINT0POINT2`, `AAUDIO_CHANNEL_3POINT1POINT2`, `AAUDIO_CHANNEL_QUAD`, `AAUDIO_CHANNEL_QUAD_SIDE`, `AAUDIO_CHANNEL_SURROUND`, `AAUDIO_CHANNEL_PENTA`, `AAUDIO_CHANNEL_5POINT1`, `AAUDIO_CHANNEL_5POINT1_SIDE`, `AAUDIO_CHANNEL_5POINT1POINT2`, `AAUDIO_CHANNEL_5POINT1POINT4`, `AAUDIO_CHANNEL_6POINT1`, `AAUDIO_CHANNEL_7POINT1`, `AAUDIO_CHANNEL_7POINT1POINT2`, `AAUDIO_CHANNEL_7POINT1POINT4`, `AAUDIO_CHANNEL_9POINT1POINT4`, `AAUDIO_CHANNEL_9POINT1POINT6` | Value obtained from FuzzedDataProvider |
 | `Direction` | `AAUDIO_DIRECTION_OUTPUT`, `AAUDIO_DIRECTION_INPUT` | Value chosen from valid values by obtaining index from FuzzedDataProvider |
 | `SharingMode` | `AAUDIO_SHARING_MODE_EXCLUSIVE`, `AAUDIO_SHARING_MODE_SHARED` | Value chosen from valid values by obtaining index from FuzzedDataProvider |
 | `Usage` | `AAUDIO_USAGE_MEDIA`, `AAUDIO_USAGE_VOICE_COMMUNICATION`, `AAUDIO_USAGE_VOICE_COMMUNICATION_SIGNALLING`, `AAUDIO_USAGE_ALARM`, `AAUDIO_USAGE_NOTIFICATION`, `AAUDIO_USAGE_NOTIFICATION_RINGTONE`, `AAUDIO_USAGE_NOTIFICATION_EVENT`, `AAUDIO_USAGE_ASSISTANCE_ACCESSIBILITY`, `AAUDIO_USAGE_ASSISTANCE_NAVIGATION_GUIDANCE`, `AAUDIO_USAGE_ASSISTANCE_SONIFICATION`, `AAUDIO_USAGE_GAME`, `AAUDIO_USAGE_ASSISTANT`, `AAUDIO_SYSTEM_USAGE_EMERGENCY`, `AAUDIO_SYSTEM_USAGE_SAFETY`, `AAUDIO_SYSTEM_USAGE_VEHICLE_STATUS`, `AAUDIO_SYSTEM_USAGE_ANNOUNCEMENT` | Value chosen from valid values by obtaining index from FuzzedDataProvider |
diff --git a/services/oboeservice/fuzzer/oboeservice_fuzzer.cpp b/services/oboeservice/fuzzer/oboeservice_fuzzer.cpp
index 4bc661c..17e8d36 100644
--- a/services/oboeservice/fuzzer/oboeservice_fuzzer.cpp
+++ b/services/oboeservice/fuzzer/oboeservice_fuzzer.cpp
@@ -68,10 +68,71 @@
     AAUDIO_INPUT_PRESET_UNPROCESSED,       AAUDIO_INPUT_PRESET_VOICE_PERFORMANCE,
 };
 
+aaudio_channel_mask_t kAAudioChannelMasks[] = {
+    AAUDIO_UNSPECIFIED,
+    AAUDIO_CHANNEL_INDEX_MASK_1,
+    AAUDIO_CHANNEL_INDEX_MASK_2,
+    AAUDIO_CHANNEL_INDEX_MASK_3,
+    AAUDIO_CHANNEL_INDEX_MASK_4,
+    AAUDIO_CHANNEL_INDEX_MASK_5,
+    AAUDIO_CHANNEL_INDEX_MASK_6,
+    AAUDIO_CHANNEL_INDEX_MASK_7,
+    AAUDIO_CHANNEL_INDEX_MASK_8,
+    AAUDIO_CHANNEL_INDEX_MASK_9,
+    AAUDIO_CHANNEL_INDEX_MASK_10,
+    AAUDIO_CHANNEL_INDEX_MASK_11,
+    AAUDIO_CHANNEL_INDEX_MASK_12,
+    AAUDIO_CHANNEL_INDEX_MASK_13,
+    AAUDIO_CHANNEL_INDEX_MASK_14,
+    AAUDIO_CHANNEL_INDEX_MASK_15,
+    AAUDIO_CHANNEL_INDEX_MASK_16,
+    AAUDIO_CHANNEL_INDEX_MASK_17,
+    AAUDIO_CHANNEL_INDEX_MASK_18,
+    AAUDIO_CHANNEL_INDEX_MASK_19,
+    AAUDIO_CHANNEL_INDEX_MASK_20,
+    AAUDIO_CHANNEL_INDEX_MASK_21,
+    AAUDIO_CHANNEL_INDEX_MASK_22,
+    AAUDIO_CHANNEL_INDEX_MASK_23,
+    AAUDIO_CHANNEL_INDEX_MASK_24,
+    AAUDIO_CHANNEL_MONO,
+    AAUDIO_CHANNEL_STEREO,
+    AAUDIO_CHANNEL_FRONT_BACK,
+    AAUDIO_CHANNEL_2POINT0POINT2,
+    AAUDIO_CHANNEL_2POINT1POINT2,
+    AAUDIO_CHANNEL_3POINT0POINT2,
+    AAUDIO_CHANNEL_3POINT1POINT2,
+    AAUDIO_CHANNEL_5POINT1,
+    AAUDIO_CHANNEL_MONO,
+    AAUDIO_CHANNEL_STEREO,
+    AAUDIO_CHANNEL_2POINT1,
+    AAUDIO_CHANNEL_TRI,
+    AAUDIO_CHANNEL_TRI_BACK,
+    AAUDIO_CHANNEL_3POINT1,
+    AAUDIO_CHANNEL_2POINT0POINT2,
+    AAUDIO_CHANNEL_2POINT1POINT2,
+    AAUDIO_CHANNEL_3POINT0POINT2,
+    AAUDIO_CHANNEL_3POINT1POINT2,
+    AAUDIO_CHANNEL_QUAD,
+    AAUDIO_CHANNEL_QUAD_SIDE,
+    AAUDIO_CHANNEL_SURROUND,
+    AAUDIO_CHANNEL_PENTA,
+    AAUDIO_CHANNEL_5POINT1,
+    AAUDIO_CHANNEL_5POINT1_SIDE,
+    AAUDIO_CHANNEL_5POINT1POINT2,
+    AAUDIO_CHANNEL_5POINT1POINT4,
+    AAUDIO_CHANNEL_6POINT1,
+    AAUDIO_CHANNEL_7POINT1,
+    AAUDIO_CHANNEL_7POINT1POINT2,
+    AAUDIO_CHANNEL_7POINT1POINT4,
+    AAUDIO_CHANNEL_9POINT1POINT4,
+    AAUDIO_CHANNEL_9POINT1POINT6,
+};
+
 const size_t kNumAAudioFormats = std::size(kAAudioFormats);
 const size_t kNumAAudioUsages = std::size(kAAudioUsages);
 const size_t kNumAAudioContentTypes = std::size(kAAudioContentTypes);
 const size_t kNumAAudioInputPresets = std::size(kAAudioInputPresets);
+const size_t kNumAAudioChannelMasks = std::size(kAAudioChannelMasks);
 
 class FuzzAAudioClient : public virtual RefBase, public AAudioServiceInterface {
    public:
@@ -305,7 +366,11 @@
 
     request.getConfiguration().setDeviceId(fdp.ConsumeIntegral<int32_t>());
     request.getConfiguration().setSampleRate(fdp.ConsumeIntegral<int32_t>());
-    request.getConfiguration().setSamplesPerFrame(fdp.ConsumeIntegral<int32_t>());
+    request.getConfiguration().setChannelMask((aaudio_channel_mask_t)(
+        fdp.ConsumeBool()
+            ? fdp.ConsumeIntegral<int32_t>()
+            : kAAudioChannelMasks[fdp.ConsumeIntegralInRange<int32_t>(
+                    0, kNumAAudioChannelMasks - 1)]));
     request.getConfiguration().setDirection(
         fdp.ConsumeBool() ? fdp.ConsumeIntegral<int32_t>()
                           : (fdp.ConsumeBool() ? AAUDIO_DIRECTION_OUTPUT : AAUDIO_DIRECTION_INPUT));