diff --git a/services/mediametrics/AudioPowerUsage.cpp b/services/mediametrics/AudioPowerUsage.cpp
new file mode 100644
index 0000000..e311bc8
--- /dev/null
+++ b/services/mediametrics/AudioPowerUsage.cpp
@@ -0,0 +1,396 @@
+/*
+ * Copyright (C) 2020 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 "AudioPowerUsage"
+#include <utils/Log.h>
+
+#include "AudioAnalytics.h"
+#include "MediaMetricsService.h"
+#include <map>
+#include <sstream>
+#include <string>
+#include <audio_utils/clock.h>
+#include <cutils/properties.h>
+#include <statslog.h>
+#include <sys/timerfd.h>
+#include <system/audio-base.h>
+
+// property to disable audio power use metrics feature, default is enabled
+#define PROP_AUDIO_METRICS_DISABLED "persist.media.audio_metrics.power_usage_disabled"
+#define AUDIO_METRICS_DISABLED_DEFAULT (false)
+
+// property to set how long to send audio power use metrics data to westworld, default is 24hrs
+#define PROP_AUDIO_METRICS_INTERVAL_HR "persist.media.audio_metrics.interval_hr"
+#define INTERVAL_HR_DEFAULT (24)
+
+// for Audio Power Usage Metrics
+#define AUDIO_POWER_USAGE_KEY_AUDIO_USAGE     "audio.power.usage"
+
+#define AUDIO_POWER_USAGE_PROP_DEVICE         "device"     // int32
+#define AUDIO_POWER_USAGE_PROP_DURATION_NS    "durationNs" // int64
+#define AUDIO_POWER_USAGE_PROP_TYPE           "type"       // int32
+#define AUDIO_POWER_USAGE_PROP_VOLUME         "volume"     // double
+
+namespace android::mediametrics {
+
+/* static */
+bool AudioPowerUsage::typeFromString(const std::string& type_string, int32_t& type) {
+    static std::map<std::string, int32_t> typeTable = {
+        { "AUDIO_STREAM_VOICE_CALL",          VOIP_CALL_TYPE },
+        { "AUDIO_STREAM_SYSTEM",              MEDIA_TYPE },
+        { "AUDIO_STREAM_RING",                RINGTONE_NOTIFICATION_TYPE },
+        { "AUDIO_STREAM_MUSIC",               MEDIA_TYPE },
+        { "AUDIO_STREAM_ALARM",               ALARM_TYPE },
+        { "AUDIO_STREAM_NOTIFICATION",        RINGTONE_NOTIFICATION_TYPE },
+
+        { "AUDIO_CONTENT_TYPE_SPEECH",        VOIP_CALL_TYPE },
+        { "AUDIO_CONTENT_TYPE_MUSIC",         MEDIA_TYPE },
+        { "AUDIO_CONTENT_TYPE_MOVIE",         MEDIA_TYPE },
+        { "AUDIO_CONTENT_TYPE_SONIFICATION",  RINGTONE_NOTIFICATION_TYPE },
+
+        { "AUDIO_USAGE_MEDIA",                MEDIA_TYPE },
+        { "AUDIO_USAGE_VOICE_COMMUNICATION",  VOIP_CALL_TYPE },
+        { "AUDIO_USAGE_ALARM",                ALARM_TYPE },
+        { "AUDIO_USAGE_NOTIFICATION",         RINGTONE_NOTIFICATION_TYPE },
+
+        { "AUDIO_SOURCE_CAMCORDER",           CAMCORDER_TYPE },
+        { "AUDIO_SOURCE_VOICE_COMMUNICATION", VOIP_CALL_TYPE },
+        { "AUDIO_SOURCE_DEFAULT",             RECORD_TYPE },
+        { "AUDIO_SOURCE_MIC",                 RECORD_TYPE },
+        { "AUDIO_SOURCE_UNPROCESSED",         RECORD_TYPE },
+        { "AUDIO_SOURCE_VOICE_RECOGNITION",   RECORD_TYPE },
+    };
+
+    auto it = typeTable.find(type_string);
+    if (it == typeTable.end()) {
+        type = UNKNOWN_TYPE;
+        return false;
+    }
+
+    type = it->second;
+    return true;
+}
+
+/* static */
+bool AudioPowerUsage::deviceFromString(const std::string& device_string, int32_t& device) {
+    static std::map<std::string, int32_t> deviceTable = {
+        { "AUDIO_DEVICE_OUT_EARPIECE",             OUTPUT_EARPIECE },
+        { "AUDIO_DEVICE_OUT_SPEAKER_SAFE",         OUTPUT_SPEAKER_SAFE },
+        { "AUDIO_DEVICE_OUT_SPEAKER",              OUTPUT_SPEAKER },
+        { "AUDIO_DEVICE_OUT_WIRED_HEADSET",        OUTPUT_WIRED_HEADSET },
+        { "AUDIO_DEVICE_OUT_WIRED_HEADPHONE",      OUTPUT_WIRED_HEADSET },
+        { "AUDIO_DEVICE_OUT_BLUETOOTH_SCO",        OUTPUT_BLUETOOTH_SCO },
+        { "AUDIO_DEVICE_OUT_BLUETOOTH_A2DP",       OUTPUT_BLUETOOTH_A2DP },
+        { "AUDIO_DEVICE_OUT_USB_HEADSET",          OUTPUT_USB_HEADSET },
+        { "AUDIO_DEVICE_OUT_BLUETOOTH_SCO_HEADSET", OUTPUT_BLUETOOTH_SCO },
+
+        { "AUDIO_DEVICE_IN_BUILTIN_MIC",           INPUT_BUILTIN_MIC },
+        { "AUDIO_DEVICE_IN_BLUETOOTH_SCO_HEADSET", INPUT_BLUETOOTH_SCO },
+        { "AUDIO_DEVICE_IN_WIRED_HEADSET",         INPUT_WIRED_HEADSET_MIC },
+        { "AUDIO_DEVICE_IN_USB_DEVICE",            INPUT_USB_HEADSET_MIC },
+        { "AUDIO_DEVICE_IN_BACK_MIC",              INPUT_BUILTIN_BACK_MIC },
+    };
+
+    auto it = deviceTable.find(device_string);
+    if (it == deviceTable.end()) {
+        device = 0;
+        return false;
+    }
+
+    device = it->second;
+    return true;
+}
+
+int32_t AudioPowerUsage::deviceFromStringPairs(const std::string& device_strings) {
+    int32_t deviceMask = 0;
+    const auto devaddrvec = MediaMetricsService::getDeviceAddressPairs(device_strings);
+    for (const auto &[device, addr] : devaddrvec) {
+        int32_t combo_device = 0;
+        deviceFromString(device, combo_device);
+        deviceMask |= combo_device;
+    }
+    return deviceMask;
+}
+
+/* static */
+void AudioPowerUsage::sendItem(const std::shared_ptr<const mediametrics::Item>& item)
+{
+    int32_t type;
+    if (!item->getInt32(AUDIO_POWER_USAGE_PROP_TYPE, &type)) return;
+
+    int32_t device;
+    if (!item->getInt32(AUDIO_POWER_USAGE_PROP_DEVICE, &device)) return;
+
+    int64_t duration_ns;
+    if (!item->getInt64(AUDIO_POWER_USAGE_PROP_DURATION_NS, &duration_ns)) return;
+
+    double volume;
+    if (!item->getDouble(AUDIO_POWER_USAGE_PROP_VOLUME, &volume)) return;
+
+#ifdef STATSD
+    (void)android::util::stats_write(android::util::AUDIO_POWER_USAGE_DATA_REPORTED,
+                                         device,
+                                         (int32_t)(duration_ns / NANOS_PER_SECOND),
+                                         (float)volume,
+                                         type);
+#endif
+}
+
+bool AudioPowerUsage::saveAsItem_l(
+        int32_t device, int64_t duration_ns, int32_t type, double average_vol)
+{
+    ALOGV("%s: (%#x, %d, %lld, %f)", __func__, device, type,
+                                   (long long)duration_ns, average_vol );
+    if (duration_ns == 0) {
+        return true; // skip duration 0 usage
+    }
+    if (device == 0) {
+        return true; //ignore unknown device
+    }
+
+    for (auto item : mItems) {
+        int32_t item_type = 0, item_device = 0;
+        double item_volume = 0.;
+        int64_t item_duration_ns = 0;
+        item->getInt32(AUDIO_POWER_USAGE_PROP_DEVICE, &item_device);
+        item->getInt64(AUDIO_POWER_USAGE_PROP_DURATION_NS, &item_duration_ns);
+        item->getInt32(AUDIO_POWER_USAGE_PROP_TYPE, &item_type);
+        item->getDouble(AUDIO_POWER_USAGE_PROP_VOLUME, &item_volume);
+
+        // aggregate by device and type
+        if (item_device == device && item_type == type) {
+            int64_t final_duration_ns = item_duration_ns + duration_ns;
+            double final_volume = (device & INPUT_DEVICE_BIT) ? 1.0:
+                            ((item_volume * item_duration_ns +
+                            average_vol * duration_ns) / final_duration_ns);
+
+            item->setInt64(AUDIO_POWER_USAGE_PROP_DURATION_NS, final_duration_ns);
+            item->setDouble(AUDIO_POWER_USAGE_PROP_VOLUME, final_volume);
+            item->setTimestamp(systemTime(SYSTEM_TIME_REALTIME));
+
+            ALOGV("%s: update (%#x, %d, %lld, %f) --> (%lld, %f)", __func__,
+                  device, type,
+                  (long long)item_duration_ns, item_volume,
+                  (long long)final_duration_ns, final_volume);
+
+            return true;
+        }
+    }
+
+    auto sitem = std::make_shared<mediametrics::Item>(AUDIO_POWER_USAGE_KEY_AUDIO_USAGE);
+    sitem->setTimestamp(systemTime(SYSTEM_TIME_REALTIME));
+    sitem->setInt32(AUDIO_POWER_USAGE_PROP_DEVICE, device);
+    sitem->setInt64(AUDIO_POWER_USAGE_PROP_DURATION_NS, duration_ns);
+    sitem->setInt32(AUDIO_POWER_USAGE_PROP_TYPE, type);
+    sitem->setDouble(AUDIO_POWER_USAGE_PROP_VOLUME, average_vol);
+    mItems.emplace_back(sitem);
+    return true;
+}
+
+void AudioPowerUsage::checkTrackRecord(
+        const std::shared_ptr<const mediametrics::Item>& item, bool isTrack)
+{
+    const std::string key = item->getKey();
+
+    int64_t deviceTimeNs = 0;
+    if (!item->getInt64(AMEDIAMETRICS_PROP_DEVICETIMENS, &deviceTimeNs)) {
+        return;
+    }
+    double deviceVolume = 1.;
+    if (isTrack && !item->getDouble(AMEDIAMETRICS_PROP_DEVICEVOLUME, &deviceVolume)) {
+        return;
+    }
+    int32_t type = 0;
+    std::string type_string;
+    if ((isTrack && mAudioAnalytics->mAnalyticsState->timeMachine().get(
+               key, AMEDIAMETRICS_PROP_STREAMTYPE, &type_string) == OK) ||
+        (!isTrack && mAudioAnalytics->mAnalyticsState->timeMachine().get(
+               key, AMEDIAMETRICS_PROP_SOURCE, &type_string) == OK)) {
+        typeFromString(type_string, type);
+
+        if (isTrack && type == UNKNOWN_TYPE &&
+                   mAudioAnalytics->mAnalyticsState->timeMachine().get(
+                   key, AMEDIAMETRICS_PROP_USAGE, &type_string) == OK) {
+            typeFromString(type_string, type);
+        }
+        if (isTrack && type == UNKNOWN_TYPE &&
+                   mAudioAnalytics->mAnalyticsState->timeMachine().get(
+                   key, AMEDIAMETRICS_PROP_CONTENTTYPE, &type_string) == OK) {
+            typeFromString(type_string, type);
+        }
+        ALOGV("type = %s => %d", type_string.c_str(), type);
+    }
+
+    int32_t device = 0;
+    std::string device_strings;
+    if ((isTrack && mAudioAnalytics->mAnalyticsState->timeMachine().get(
+         key, AMEDIAMETRICS_PROP_OUTPUTDEVICES, &device_strings) == OK) ||
+        (!isTrack && mAudioAnalytics->mAnalyticsState->timeMachine().get(
+         key, AMEDIAMETRICS_PROP_INPUTDEVICES, &device_strings) == OK)) {
+
+        device = deviceFromStringPairs(device_strings);
+        ALOGV("device = %s => %d", device_strings.c_str(), device);
+    }
+    std::lock_guard l(mLock);
+    saveAsItem_l(device, deviceTimeNs, type, deviceVolume);
+}
+
+void AudioPowerUsage::checkMode(const std::shared_ptr<const mediametrics::Item>& item)
+{
+    std::string mode;
+    if (!item->getString(AMEDIAMETRICS_PROP_AUDIOMODE, &mode)) return;
+
+    std::lock_guard l(mLock);
+    if (mode == mMode) return;  // no change in mode.
+
+    if (mMode == "AUDIO_MODE_IN_CALL") { // leaving call mode
+        const int64_t endCallNs = item->getTimestamp();
+        const int64_t durationNs = endCallNs - mDeviceTimeNs;
+        if (durationNs > 0) {
+            mDeviceVolume = (mDeviceVolume * (mVolumeTimeNs - mDeviceTimeNs) +
+                    mVoiceVolume * (endCallNs - mVolumeTimeNs)) / durationNs;
+            saveAsItem_l(mPrimaryDevice, durationNs, VOICE_CALL_TYPE, mDeviceVolume);
+        }
+    } else if (mode == "AUDIO_MODE_IN_CALL") { // entering call mode
+        mStartCallNs = item->getTimestamp(); // advisory only
+
+        mDeviceVolume = 0;
+        mVolumeTimeNs = mStartCallNs;
+        mDeviceTimeNs = mStartCallNs;
+    }
+    ALOGV("%s: new mode:%s  old mode:%s", __func__, mode.c_str(), mMode.c_str());
+    mMode = mode;
+}
+
+void AudioPowerUsage::checkVoiceVolume(const std::shared_ptr<const mediametrics::Item>& item)
+{
+    double voiceVolume = 0.;
+    if (!item->getDouble(AMEDIAMETRICS_PROP_VOICEVOLUME, &voiceVolume)) return;
+
+    std::lock_guard l(mLock);
+    if (voiceVolume == mVoiceVolume) return;  // no change in volume
+
+    // we only track average device volume when we are in-call
+    if (mMode == "AUDIO_MODE_IN_CALL") {
+        const int64_t timeNs = item->getTimestamp();
+        const int64_t durationNs = timeNs - mDeviceTimeNs;
+        if (durationNs > 0) {
+            mDeviceVolume = (mDeviceVolume * (mVolumeTimeNs - mDeviceTimeNs) +
+                    mVoiceVolume * (timeNs - mVolumeTimeNs)) / durationNs;
+            mVolumeTimeNs = timeNs;
+        }
+    }
+    ALOGV("%s: new voice volume:%lf  old voice volume:%lf", __func__, voiceVolume, mVoiceVolume);
+    mVoiceVolume = voiceVolume;
+}
+
+void AudioPowerUsage::checkCreatePatch(const std::shared_ptr<const mediametrics::Item>& item)
+{
+    std::string outputDevices;
+    if (!item->get(AMEDIAMETRICS_PROP_OUTPUTDEVICES, &outputDevices)) return;
+
+    const std::string& key = item->getKey();
+    std::string flags;
+    if (mAudioAnalytics->mAnalyticsState->timeMachine().get(
+         key, AMEDIAMETRICS_PROP_FLAGS, &flags) != OK) return;
+
+    if (flags.find("AUDIO_OUTPUT_FLAG_PRIMARY") == std::string::npos) return;
+
+    const int32_t device = deviceFromStringPairs(outputDevices);
+
+    std::lock_guard l(mLock);
+    if (mPrimaryDevice == device) return;
+
+    if (mMode == "AUDIO_MODE_IN_CALL") {
+        // Save statistics
+        const int64_t endDeviceNs = item->getTimestamp();
+        const int64_t durationNs = endDeviceNs - mDeviceTimeNs;
+        if (durationNs > 0) {
+            mDeviceVolume = (mDeviceVolume * (mVolumeTimeNs - mDeviceTimeNs) +
+                    mVoiceVolume * (endDeviceNs - mVolumeTimeNs)) / durationNs;
+            saveAsItem_l(mPrimaryDevice, durationNs, VOICE_CALL_TYPE, mDeviceVolume);
+        }
+        // reset statistics
+        mDeviceVolume = 0;
+        mDeviceTimeNs = endDeviceNs;
+        mVolumeTimeNs = endDeviceNs;
+    }
+    ALOGV("%s: new primary device:%#x  old primary device:%#x", __func__, device, mPrimaryDevice);
+    mPrimaryDevice = device;
+}
+
+AudioPowerUsage::AudioPowerUsage(AudioAnalytics *audioAnalytics)
+    : mAudioAnalytics(audioAnalytics)
+    , mDisabled(property_get_bool(PROP_AUDIO_METRICS_DISABLED, AUDIO_METRICS_DISABLED_DEFAULT))
+    , mIntervalHours(property_get_int32(PROP_AUDIO_METRICS_INTERVAL_HR, INTERVAL_HR_DEFAULT))
+{
+    ALOGD("%s", __func__);
+    ALOGI_IF(mDisabled, "AudioPowerUsage is disabled.");
+    collect(); // send items
+}
+
+AudioPowerUsage::~AudioPowerUsage()
+{
+    ALOGD("%s", __func__);
+}
+
+void AudioPowerUsage::clear()
+{
+    std::lock_guard _l(mLock);
+    mItems.clear();
+}
+
+void AudioPowerUsage::collect()
+{
+    std::lock_guard _l(mLock);
+    for (const auto &item : mItems) {
+        sendItem(item);
+    }
+    mItems.clear();
+    mAudioAnalytics->mTimedAction.postIn(
+        mIntervalHours <= 0 ? std::chrono::seconds(5) : std::chrono::hours(mIntervalHours),
+        [this](){ collect(); });
+}
+
+std::pair<std::string, int32_t> AudioPowerUsage::dump(int limit) const {
+    if (limit <= 2) {
+        return {{}, 0};
+    }
+    std::lock_guard _l(mLock);
+    if (mDisabled) {
+        return {"AudioPowerUsage disabled\n", 1};
+    }
+    if (mItems.empty()) {
+        return {"AudioPowerUsage empty\n", 1};
+    }
+
+    int slot = 1;
+    std::stringstream ss;
+    ss << "AudioPowerUsage:\n";
+    for (const auto &item : mItems) {
+        if (slot >= limit - 1) {
+            ss << "-- AudioPowerUsage may be truncated!\n";
+            ++slot;
+            break;
+        }
+        ss << " " << slot << " " << item->toString() << "\n";
+        slot++;
+    }
+    return { ss.str(), slot };
+}
+
+} // namespace android