fixed bug in Audio.cpp where numSilentSamples was written to wrong place in packet

added more debug stuff, other minor changes and fixes

assignment-client/src/audio/AudioMixerClientData.cpp

@@ -119,15 +119,14 @@ void AudioMixerClientData::checkBuffersBeforeFrameSend(AABox* checkSourceZone, A
 }
 
 void AudioMixerClientData::pushBuffersAfterFrameSend() {
+
     QList<PositionalAudioRingBuffer*>::iterator i = _ringBuffers.begin();
     while (i != _ringBuffers.end()) {
         // this was a used buffer, push the output pointer forwards
         PositionalAudioRingBuffer* audioBuffer = *i;
 
         if (audioBuffer->willBeAddedToMix()) {
-            audioBuffer->shiftReadPosition(audioBuffer->isStereo()
+            audioBuffer->shiftReadPosition(audioBuffer->getSamplesPerFrame());
-                                           ? NETWORK_BUFFER_LENGTH_SAMPLES_STEREO : NETWORK_BUFFER_LENGTH_SAMPLES_PER_CHANNEL);
-
             audioBuffer->setWillBeAddedToMix(false);
         } else if (audioBuffer->getType() == PositionalAudioRingBuffer::Injector
                    && audioBuffer->hasStarted() && audioBuffer->isStarved()) {

assignment-client/src/audio/AvatarAudioRingBuffer.cpp

@@ -20,7 +20,7 @@ AvatarAudioRingBuffer::AvatarAudioRingBuffer(bool isStereo) :
 
 int AvatarAudioRingBuffer::parseData(const QByteArray& packet) {
     _interframeTimeGapHistory.frameReceived();
-    updateDesiredJitterBufferNumSamples();
+    updateDesiredJitterBufferFrames();
 
     _shouldLoopbackForNode = (packetTypeForPacket(packet) == PacketTypeMicrophoneAudioWithEcho);
     return PositionalAudioRingBuffer::parseData(packet);

interface/src/Audio.cpp

@@ -461,8 +461,8 @@ void Audio::handleAudioInput() {
         int16_t* inputAudioSamples = new int16_t[inputSamplesRequired];
         _inputRingBuffer.readSamples(inputAudioSamples, inputSamplesRequired);
         
-        int numNetworkBytes = _isStereoInput ? NETWORK_BUFFER_LENGTH_BYTES_STEREO : NETWORK_BUFFER_LENGTH_BYTES_PER_CHANNEL;
+        const int numNetworkBytes = _isStereoInput ? NETWORK_BUFFER_LENGTH_BYTES_STEREO : NETWORK_BUFFER_LENGTH_BYTES_PER_CHANNEL;
-        int numNetworkSamples = _isStereoInput ? NETWORK_BUFFER_LENGTH_SAMPLES_STEREO : NETWORK_BUFFER_LENGTH_SAMPLES_PER_CHANNEL;
+        const int numNetworkSamples = _isStereoInput ? NETWORK_BUFFER_LENGTH_SAMPLES_STEREO : NETWORK_BUFFER_LENGTH_SAMPLES_PER_CHANNEL;
 
         // zero out the monoAudioSamples array and the locally injected audio
         memset(networkAudioSamples, 0, numNetworkBytes);
@@ -622,6 +622,7 @@ void Audio::handleAudioInput() {
         SharedNodePointer audioMixer = nodeList->soloNodeOfType(NodeType::AudioMixer);
         
         if (audioMixer && audioMixer->getActiveSocket()) {
+
             MyAvatar* interfaceAvatar = Application::getInstance()->getAvatar();
             glm::vec3 headPosition = interfaceAvatar->getHead()->getPosition();
             glm::quat headOrientation = interfaceAvatar->getHead()->getFinalOrientationInWorldFrame();
@@ -634,12 +635,11 @@ void Audio::handleAudioInput() {
                 packetType = PacketTypeSilentAudioFrame;
                 
                 // we need to indicate how many silent samples this is to the audio mixer
-                audioDataPacket[0] = _isStereoInput
+                networkAudioSamples[0] = numNetworkSamples;
-                    ? NETWORK_BUFFER_LENGTH_SAMPLES_STEREO
-                    : NETWORK_BUFFER_LENGTH_SAMPLES_PER_CHANNEL;
                 numAudioBytes = sizeof(int16_t);
             } else {
-                numAudioBytes = _isStereoInput ? NETWORK_BUFFER_LENGTH_BYTES_STEREO : NETWORK_BUFFER_LENGTH_BYTES_PER_CHANNEL;
+                numAudioBytes = numNetworkBytes;
+                //_isStereoInput ? NETWORK_BUFFER_LENGTH_BYTES_STEREO : NETWORK_BUFFER_LENGTH_BYTES_PER_CHANNEL;
                 
                 if (Menu::getInstance()->isOptionChecked(MenuOption::EchoServerAudio)) {
                     packetType = PacketTypeMicrophoneAudioWithEcho;

libraries/audio/src/AudioRingBuffer.cpp

@@ -158,6 +158,7 @@ const int16_t& AudioRingBuffer::operator[] (const int index) const {
 
 void AudioRingBuffer::shiftReadPosition(unsigned int numSamples) {
     _nextOutput = shiftedPositionAccomodatingWrap(_nextOutput, numSamples);
+    printf("\n mixed.  %d samples remaining\n", samplesAvailable());
 }
 
 unsigned int AudioRingBuffer::samplesAvailable() const {

libraries/audio/src/InjectedAudioRingBuffer.cpp

@@ -32,7 +32,7 @@ const uchar MAX_INJECTOR_VOLUME = 255;
 
 int InjectedAudioRingBuffer::parseData(const QByteArray& packet) {
     _interframeTimeGapHistory.frameReceived();
-    updateDesiredJitterBufferNumSamples();
+    updateDesiredJitterBufferFrames();
 
     // setup a data stream to read from this packet
     QDataStream packetStream(packet);

libraries/audio/src/PositionalAudioRingBuffer.cpp

@@ -33,6 +33,10 @@ InterframeTimeGapHistory::InterframeTimeGapHistory()
 }
 
 void InterframeTimeGapHistory::frameReceived() {
+
+    static QQueue<quint64> gaps;
+    static quint64 gapsSum = 0;
+
     quint64 now = usecTimestampNow();
 
     // make sure this isn't the first time frameReceived() is called, meaning there's actually a gap to calculate.
@@ -87,9 +91,9 @@ PositionalAudioRingBuffer::PositionalAudioRingBuffer(PositionalAudioRingBuffer::
     _shouldOutputStarveDebug(true),
     _isStereo(isStereo),
     _listenerUnattenuatedZone(NULL),
-    _desiredJitterBufferNumSamples(getNumSamplesPerFrame())
+    _desiredJitterBufferFrames(1),
+    _currentJitterBufferFrames(0)
 {
-
 }
 
 int PositionalAudioRingBuffer::parseData(const QByteArray& packet) {
@@ -114,10 +118,15 @@ int PositionalAudioRingBuffer::parseData(const QByteArray& packet) {
         if (numSilentSamples > 0) {
             addSilentFrame(numSilentSamples);
         }
+        printf("\nparsed silent packet of %d samples\n", numSilentSamples);
     } else {
         // there is audio data to read
-        readBytes += writeData(packet.data() + readBytes, packet.size() - readBytes);
+        int dataBytes = writeData(packet.data() + readBytes, packet.size() - readBytes);
+        readBytes += dataBytes;
+
+        printf("\nparsed packet of %d data bytes\n", dataBytes);
     }
+    printf("%d samples available\n", samplesAvailable());
     
     return readBytes;
 }
@@ -166,40 +175,58 @@ void PositionalAudioRingBuffer::updateNextOutputTrailingLoudness() {
 
 bool PositionalAudioRingBuffer::shouldBeAddedToMix() {
 
-    int samplesPerFrame = getNumSamplesPerFrame();
+    int samplesPerFrame = getSamplesPerFrame();
+    int currentJitterBufferSamples = 3 * samplesPerFrame; //_currentJitterBufferFrames * samplesPerFrame;
+    
+//printf("\nsamples available: %d  frames available: %d\n", samplesAvailable(), samplesAvailable() / samplesPerFrame);
+    if (!isNotStarvedOrHasMinimumSamples(samplesPerFrame + currentJitterBufferSamples)) {
 
-    if (!isNotStarvedOrHasMinimumSamples(samplesPerFrame + _desiredJitterBufferNumSamples)) {
+//printf("\nMIXING DELAYED! waiting for jitter buffer to fill after being starved\n");
+//printf("samples available: %d  frames available: %d\n", samplesAvailable(), samplesAvailable() / samplesPerFrame);
+        // if the buffer was starved and hasn't filled back up all the way, don't mix yet
         if (_shouldOutputStarveDebug) {
             _shouldOutputStarveDebug = false;
         }
-        
+
-        return false;
+        return  false;
+
     } else if (samplesAvailable() < samplesPerFrame) {
+
+//printf("\nMIXING DELAYED! jitter buffer is starved!!!\n");
+//printf("samples available: %d  frames available: %d\n", samplesAvailable(), samplesAvailable() / samplesPerFrame);
+        // if the buffer doesn't have a full frame of samples for mixing, it is starved
         _isStarved = true;
         
         // reset our _shouldOutputStarveDebug to true so the next is printed
         _shouldOutputStarveDebug = true;
+
+        // if buffer was starved, we've effectively increased the jitter buffer by one frame
+        // by "holding back" this ring buffer's contents until the next client frame is prepared.
+        _currentJitterBufferFrames++;
+//printf("jbuffer size increased: new size: %d\n", _currentJitterBufferFrames);
         
         return false;
-    } else {
-        // good buffer, add this to the mix
-        _isStarved = false;
 
-        // since we've read data from ring buffer at least once - we've started
-        _hasStarted = true;
-
-        return true;
     }
+//printf("WILL MIX\n");
 
-    return false;
+    // good buffer, add this to the mix
+    _isStarved = false;
+
+    // since we've read data from ring buffer at least once - we've started
+    _hasStarted = true;
+
+    return true;
 }
 
-void PositionalAudioRingBuffer::updateDesiredJitterBufferNumSamples() {
+void PositionalAudioRingBuffer::updateDesiredJitterBufferFrames() {
 
     const float USECS_PER_FRAME = NETWORK_BUFFER_LENGTH_SAMPLES_PER_CHANNEL * USECS_PER_SECOND / (float)SAMPLE_RATE;
 
     if (_interframeTimeGapHistory.hasNewWindowMaxGapAvailable()) {
-        int desiredJitterBufferNumFrames = (int)((float)_interframeTimeGapHistory.getWindowMaxGap() / USECS_PER_FRAME + 0.5f);
+        _desiredJitterBufferFrames = ceilf((float)_interframeTimeGapHistory.getWindowMaxGap() / USECS_PER_FRAME);
-        _desiredJitterBufferNumSamples = desiredJitterBufferNumFrames * getNumSamplesPerFrame();
+        if (_desiredJitterBufferFrames < 1) {
+            _desiredJitterBufferFrames = 1;
+        }
     }
 }

libraries/audio/src/PositionalAudioRingBuffer.h

@@ -29,6 +29,7 @@ public:
 
     void frameReceived();
     bool hasNewWindowMaxGapAvailable() const { return _newWindowMaxGapAvailable; }
+    quint64 peekWindowMaxGap() const { return _windowMaxGap; }
     quint64 getWindowMaxGap();
 
 private:
@@ -75,14 +76,14 @@ public:
     AABox* getListenerUnattenuatedZone() const { return _listenerUnattenuatedZone; }
     void setListenerUnattenuatedZone(AABox* listenerUnattenuatedZone) { _listenerUnattenuatedZone = listenerUnattenuatedZone; }
     
-    int getNumSamplesPerFrame() const { return _isStereo ? NETWORK_BUFFER_LENGTH_SAMPLES_STEREO : NETWORK_BUFFER_LENGTH_BYTES_PER_CHANNEL; }
+    int getSamplesPerFrame() const { return _isStereo ? NETWORK_BUFFER_LENGTH_SAMPLES_STEREO : NETWORK_BUFFER_LENGTH_SAMPLES_PER_CHANNEL; }
 
 protected:
     // disallow copying of PositionalAudioRingBuffer objects
     PositionalAudioRingBuffer(const PositionalAudioRingBuffer&);
     PositionalAudioRingBuffer& operator= (const PositionalAudioRingBuffer&);
 
-    void updateDesiredJitterBufferNumSamples();
+    void updateDesiredJitterBufferFrames();
     
     PositionalAudioRingBuffer::Type _type;
     glm::vec3 _position;
@@ -96,7 +97,10 @@ protected:
     AABox* _listenerUnattenuatedZone;
 
     InterframeTimeGapHistory _interframeTimeGapHistory;
-    int _desiredJitterBufferNumSamples;
+    int _desiredJitterBufferFrames;
+    int _currentJitterBufferFrames;
+
+quint64 _lastMixTime;
 };
 
 #endif // hifi_PositionalAudioRingBuffer_h