Change the default time scale for audio/video track during recording

and reduce rounding errors in calculating the sample duration

- Default time scale for tracks other than audio is set to 90000.
- Audio track by default uses the audio sampling rate as the time scale.
- Default movie time scale remains to be 1000.
- The default time scale values will be overwritten by a user-supplied value if exits.

Change-Id: I81b40ed0626ea45e9fd24a89e21a2c5a4a2c3415

2 files changed, 63 insertions, 14 deletions

diff --git a/media/libmediaplayerservice/StagefrightRecorder.cpp b/media/libmediaplayerservice/StagefrightRecorder.cpp
index 8481d49..94448c1 100644
--- a/media/libmediaplayerservice/StagefrightRecorder.cpp
+++ b/media/libmediaplayerservice/StagefrightRecorder.cpp
@@ -675,7 +675,9 @@ sp<MediaSource> StagefrightRecorder::createAudioSource() {
     encMeta->setInt32(kKeyChannelCount, mAudioChannels);
     encMeta->setInt32(kKeySampleRate, mSampleRate);
     encMeta->setInt32(kKeyBitRate, mAudioBitRate);
-    encMeta->setInt32(kKeyTimeScale, mAudioTimeScale);
+    if (mAudioTimeScale > 0) {
+        encMeta->setInt32(kKeyTimeScale, mAudioTimeScale);
+    }
 
     OMXClient client;
     CHECK_EQ(client.connect(), OK);
@@ -961,7 +963,9 @@ status_t StagefrightRecorder::setupVideoEncoder(sp<MediaSource> *source) {
     enc_meta->setInt32(kKeyStride, stride);
     enc_meta->setInt32(kKeySliceHeight, sliceHeight);
     enc_meta->setInt32(kKeyColorFormat, colorFormat);
-    enc_meta->setInt32(kKeyTimeScale, mVideoTimeScale);
+    if (mVideoTimeScale > 0) {
+        enc_meta->setInt32(kKeyTimeScale, mVideoTimeScale);
+    }
     if (mVideoEncoderProfile != -1) {
         enc_meta->setInt32(kKeyVideoProfile, mVideoEncoderProfile);
     }
@@ -1041,7 +1045,9 @@ status_t StagefrightRecorder::startMPEG4Recording() {
     meta->setInt32(kKeyFileType, mOutputFormat);
     meta->setInt32(kKeyBitRate, totalBitRate);
     meta->setInt32(kKey64BitFileOffset, mUse64BitFileOffset);
-    meta->setInt32(kKeyTimeScale, mMovieTimeScale);
+    if (mMovieTimeScale > 0) {
+        meta->setInt32(kKeyTimeScale, mMovieTimeScale);
+    }
     if (mTrackEveryTimeDurationUs > 0) {
         meta->setInt64(kKeyTrackTimeStatus, mTrackEveryTimeDurationUs);
     }
@@ -1117,9 +1123,9 @@ status_t StagefrightRecorder::reset() {
     mIFramesIntervalSec = 1;
     mAudioSourceNode = 0;
     mUse64BitFileOffset = false;
-    mMovieTimeScale  = 1000;
-    mAudioTimeScale  = 1000;
-    mVideoTimeScale  = 1000;
+    mMovieTimeScale  = -1;
+    mAudioTimeScale  = -1;
+    mVideoTimeScale  = -1;
     mCameraId        = 0;
     mVideoEncoderProfile = -1;
     mVideoEncoderLevel   = -1;
diff --git a/media/libstagefright/MPEG4Writer.cpp b/media/libstagefright/MPEG4Writer.cpp
index 9fe3864..f52ec1a 100644
--- a/media/libstagefright/MPEG4Writer.cpp
+++ b/media/libstagefright/MPEG4Writer.cpp
@@ -168,6 +168,12 @@ private:
 
     void getCodecSpecificDataFromInputFormatIfPossible();
 
+    // Determine the track time scale
+    // If it is an audio track, try to use the sampling rate as
+    // the time scale; however, if user chooses the overwrite
+    // value, the user-supplied time scale will be used.
+    void setTimeScale();
+
     Track(const Track &);
     Track &operator=(const Track &);
 };
@@ -434,7 +440,7 @@ void MPEG4Writer::stop() {
     mMoovBoxBuffer = (uint8_t *) malloc(mEstimatedMoovBoxSize);
     mMoovBoxBufferOffset = 0;
     CHECK(mMoovBoxBuffer != NULL);
-    int32_t duration = (maxDurationUs * mTimeScale) / 1E6;
+    int32_t duration = (maxDurationUs * mTimeScale + 5E5) / 1E6;
 
     beginBox("moov");
 
@@ -749,10 +755,6 @@ MPEG4Writer::Track::Track(
       mReachedEOS(false) {
     getCodecSpecificDataFromInputFormatIfPossible();
 
-    if (!mMeta->findInt32(kKeyTimeScale, &mTimeScale)) {
-        mTimeScale = 1000;
-    }
-
     const char *mime;
     mMeta->findCString(kKeyMIMEType, &mime);
     mIsAvc = !strcasecmp(mime, MEDIA_MIMETYPE_VIDEO_AVC);
@@ -760,6 +762,28 @@ MPEG4Writer::Track::Track(
     mIsMPEG4 = !strcasecmp(mime, MEDIA_MIMETYPE_VIDEO_MPEG4) ||
                !strcasecmp(mime, MEDIA_MIMETYPE_AUDIO_AAC);
 
+    setTimeScale();
+}
+
+void MPEG4Writer::Track::setTimeScale() {
+    LOGV("setTimeScale");
+    // Default time scale
+    mTimeScale = 90000;
+
+    if (mIsAudio) {
+        // Use the sampling rate as the default time scale for audio track.
+        int32_t sampleRate;
+        bool success = mMeta->findInt32(kKeySampleRate, &sampleRate);
+        CHECK(success);
+        mTimeScale = sampleRate;
+    }
+
+    // If someone would like to overwrite the timescale, use user-supplied value.
+    int32_t timeScale;
+    if (mMeta->findInt32(kKeyTimeScale, &timeScale)) {
+        mTimeScale = timeScale;
+    }
+
     CHECK(mTimeScale > 0);
 }
 
@@ -1336,6 +1360,8 @@ void MPEG4Writer::Track::threadEntry() {
     int32_t nZeroLengthFrames = 0;
     int64_t lastTimestampUs = 0;  // Previous sample time stamp in ms
     int64_t lastDurationUs = 0;   // Between the previous two samples in ms
+    int64_t currDurationTicks = 0;  // Timescale based ticks
+    int64_t lastDurationTicks = 0;  // Timescale based ticks
     int32_t sampleCount = 1;      // Sample count in the current stts table entry
     uint32_t previousSampleSize = 0;  // Size of the previous sample
     int64_t previousPausedDurationUs = 0;
@@ -1483,7 +1509,16 @@ void MPEG4Writer::Track::threadEntry() {
         mSampleSizes.push_back(sampleSize);
         ++mNumSamples;
         if (mNumSamples > 2) {
-            if (lastDurationUs != timestampUs - lastTimestampUs) {
+            // We need to use the time scale based ticks, rather than the
+            // timestamp itself to determine whether we have to use a new
+            // stts entry, since we may have rounding errors.
+            // The calculation is intended to reduce the accumulated
+            // rounding errors.
+            currDurationTicks =
+                     ((timestampUs * mTimeScale + 500000LL) / 1000000LL -
+                     (lastTimestampUs * mTimeScale + 500000LL) / 1000000LL);
+
+            if (currDurationTicks != lastDurationTicks) {
                 SttsTableEntry sttsEntry(sampleCount, lastDurationUs);
                 mSttsTableEntries.push_back(sttsEntry);
                 sampleCount = 1;
@@ -1498,6 +1533,7 @@ void MPEG4Writer::Track::threadEntry() {
             previousSampleSize = sampleSize;
         }
         lastDurationUs = timestampUs - lastTimestampUs;
+        lastDurationTicks = currDurationTicks;
         lastTimestampUs = timestampUs;
         if (mIsRealTimeRecording && mIsAudio) {
             wallClockTimeUs = systemTime() / 1000;
@@ -1774,7 +1810,6 @@ void MPEG4Writer::Track::writeTrackHeader(
     LOGV("%s track time scale: %d",
         mIsAudio? "Audio": "Video", mTimeScale);
 
-
     time_t now = time(NULL);
     int32_t mvhdTimeScale = mOwner->getTimeScale();
     int64_t trakDurationUs = getDurationUs();
@@ -2089,10 +2124,18 @@ void MPEG4Writer::Track::writeTrackHeader(
           mOwner->beginBox("stts");
             mOwner->writeInt32(0);  // version=0, flags=0
             mOwner->writeInt32(mSttsTableEntries.size());
+            int64_t prevTimestampUs = 0;
             for (List<SttsTableEntry>::iterator it = mSttsTableEntries.begin();
                  it != mSttsTableEntries.end(); ++it) {
                 mOwner->writeInt32(it->sampleCount);
-                int32_t dur = (it->sampleDurationUs * mTimeScale + 5E5) / 1E6;
+
+                // Make sure that we are calculating the sample duration the exactly
+                // same way as we made decision on how to create stts entries.
+                int64_t currTimestampUs = prevTimestampUs + it->sampleDurationUs;
+                int32_t dur = ((currTimestampUs * mTimeScale + 500000LL) / 1000000LL -
+                               (prevTimestampUs * mTimeScale + 500000LL) / 1000000LL);
+                prevTimestampUs += (it->sampleCount * it->sampleDurationUs);
+
                 mOwner->writeInt32(dur);
             }
           mOwner->endBox();  // stts