void
AudioSink::NotifyAudioNeeded()
{
MOZ_ASSERT(mOwnerThread->IsCurrentThreadIn(),
"Not called from the owner's thread");
// Always ensure we have two processed frames pending to allow for processing
// latency.
while (mAudioQueue.GetSize() && (mAudioQueue.IsFinished() ||
mProcessedQueueLength < LOW_AUDIO_USECS ||
mProcessedQueue.GetSize() < 2)) {
RefPtr<AudioData> data = mAudioQueue.PopFront();
// Ignore the element with 0 frames and try next.
if (!data->mFrames) {
continue;
}
if (!mConverter ||
(data->mRate != mConverter->InputConfig().Rate() ||
data->mChannels != mConverter->InputConfig().Channels())) {
SINK_LOG_V("Audio format changed from %u@%uHz to %u@%uHz",
mConverter? mConverter->InputConfig().Channels() : 0,
mConverter ? mConverter->InputConfig().Rate() : 0,
data->mChannels, data->mRate);
DrainConverter();
// mFramesParsed indicates the current playtime in frames at the current
// input sampling rate. Recalculate it per the new sampling rate.
if (mFramesParsed) {
// We minimize overflow.
uint32_t oldRate = mConverter->InputConfig().Rate();
uint32_t newRate = data->mRate;
CheckedInt64 result = SaferMultDiv(mFramesParsed, newRate, oldRate);
if (!result.isValid()) {
NS_WARNING("Int overflow in AudioSink");
mErrored = true;
return;
}
mFramesParsed = result.value();
}
mConverter =
MakeUnique<AudioConverter>(
AudioConfig(data->mChannels, data->mRate),
AudioConfig(mOutputChannels, mOutputRate));
}
// See if there's a gap in the audio. If there is, push silence into the
// audio hardware, so we can play across the gap.
// Calculate the timestamp of the next chunk of audio in numbers of
// samples.
CheckedInt64 sampleTime =
TimeUnitToFrames(data->mTime - mStartTime, data->mRate);
// Calculate the number of frames that have been pushed onto the audio hardware.
CheckedInt64 missingFrames = sampleTime - mFramesParsed;
if (!missingFrames.isValid()) {
NS_WARNING("Int overflow in AudioSink");
mErrored = true;
return;
}
if (missingFrames.value() > AUDIO_FUZZ_FRAMES) {
// The next audio packet begins some time after the end of the last packet
// we pushed to the audio hardware. We must push silence into the audio
// hardware so that the next audio packet begins playback at the correct
// time.
missingFrames = std::min<int64_t>(INT32_MAX, missingFrames.value());
mFramesParsed += missingFrames.value();
RefPtr<AudioData> silenceData;
AlignedAudioBuffer silenceBuffer(missingFrames.value() * data->mChannels);
if (!silenceBuffer) {
NS_WARNING("OOM in AudioSink");
mErrored = true;
return;
}
if (mConverter->InputConfig() != mConverter->OutputConfig()) {
AlignedAudioBuffer convertedData =
mConverter->Process(AudioSampleBuffer(Move(silenceBuffer))).Forget();
silenceData = CreateAudioFromBuffer(Move(convertedData), data);
} else {
silenceData = CreateAudioFromBuffer(Move(silenceBuffer), data);
}
PushProcessedAudio(silenceData);
}
mLastEndTime = data->GetEndTime();
mFramesParsed += data->mFrames;
if (mConverter->InputConfig() != mConverter->OutputConfig()) {
// We must ensure that the size in the buffer contains exactly the number
// of frames, in case one of the audio producer over allocated the buffer.
AlignedAudioBuffer buffer(Move(data->mAudioData));
buffer.SetLength(size_t(data->mFrames) * data->mChannels);
AlignedAudioBuffer convertedData =
mConverter->Process(AudioSampleBuffer(Move(buffer))).Forget();
data = CreateAudioFromBuffer(Move(convertedData), data);
}
if (PushProcessedAudio(data)) {
mLastProcessedPacket = Some(data);
}
}
if (mAudioQueue.IsFinished()) {
// We have reached the end of the data, drain the resampler.
DrainConverter();
mProcessedQueue.Finish();
}
}
// Converts from microseconds to number of audio frames, given the specified
// audio rate.
CheckedInt64 UsecsToFrames(int64_t aUsecs, uint32_t aRate) {
return SaferMultDiv(aUsecs, aRate, USECS_PER_S);
}
// Converts from number of audio frames to microseconds, given the specified
// audio rate.
CheckedInt64 FramesToUsecs(int64_t aFrames, uint32_t aRate) {
return SaferMultDiv(aFrames, USECS_PER_S, aRate);
}
