Audio recording for podcasts on OSX

Re-coded this feature (which had been temporarily removed due to the
old code being reliant on the obsolete QuickTime library).

src/podcast/quicktime/UBAudioQueueRecorder.cpp

@@ -37,15 +37,22 @@ UBAudioQueueRecorder::UBAudioQueueRecorder(QObject* pParent)
     , mIsRecording(false)
     , mBufferLengthInMs(500)
 {
+
+    int sampleSize = sizeof(float); // TODO: check if this is system/ microphone-dependant
+    
     sAudioFormat.mSampleRate = 44100.0;
-    sAudioFormat.mFormatID = kAudioFormatMPEG4AAC;
-    sAudioFormat.mChannelsPerFrame = 2;
-
-    sAudioFormat.mBytesPerFrame = 0;
-    sAudioFormat.mBitsPerChannel = 0;
-    sAudioFormat.mBytesPerPacket = 0;
-    sAudioFormat.mFramesPerPacket = 0;
-    sAudioFormat.mFormatFlags = 0;
+    sAudioFormat.mFormatID = kAudioFormatLinearPCM;
+    sAudioFormat.mChannelsPerFrame = 1;
+
+    sAudioFormat.mBytesPerFrame = sampleSize;
+    sAudioFormat.mBitsPerChannel = 8 * sampleSize;
+    sAudioFormat.mBytesPerPacket = sampleSize;
+    sAudioFormat.mFramesPerPacket = 1;
+    sAudioFormat.mFormatFlags = kAudioFormatFlagIsFloat | 
+                                kAudioFormatFlagsNativeEndian | 
+                                kAudioFormatFlagIsPacked;
+    
+    
 }
 
 
@@ -250,7 +257,7 @@ bool UBAudioQueueRecorder::init(const QString& waveInDeviceName)
 
     int nbBuffers = 6;
     mSampleBufferSize = sAudioFormat.mSampleRate *  sAudioFormat.mChannelsPerFrame
-                * 2 * mBufferLengthInMs / 1000; // 44.1 Khz * stereo * 16bit * buffer length
+                * sAudioFormat.mChannelsPerFrame * mBufferLengthInMs / 1000; // 44.1 Khz * stereo * 16bit * buffer length
 
     for (int i = 0; i < nbBuffers; i++)
     {
@@ -333,9 +340,12 @@ bool UBAudioQueueRecorder::close()
 }
 
 
-void UBAudioQueueRecorder::audioQueueInputCallback (void *inUserData, AudioQueueRef inAQ,
-                                                            AudioQueueBufferRef inBuffer, const AudioTimeStamp *inStartTime,
-                                                            UInt32 inNumberPacketDescriptions, const AudioStreamPacketDescription *inPacketDescs)
+void UBAudioQueueRecorder::audioQueueInputCallback (void *inUserData, 
+                                                    AudioQueueRef inAQ,
+                                                    AudioQueueBufferRef inBuffer, 
+                                                    const AudioTimeStamp *inStartTime,
+                                                    UInt32 inNumberPacketDescriptions, 
+                                                    const AudioStreamPacketDescription *inPacketDescs)
 {
     Q_UNUSED(inAQ);
     Q_UNUSED(inStartTime)
@@ -356,12 +366,11 @@ void UBAudioQueueRecorder::audioQueueInputCallback (void *inUserData, AudioQueue
 
 
 void UBAudioQueueRecorder::emitNewWaveBuffer(AudioQueueBufferRef pBuffer,
-                int inNumberPacketDescriptions, const AudioStreamPacketDescription *inPacketDescs)
+                                             int inNumberPacketDescriptions, 
+                                             const AudioStreamPacketDescription *inPacketDescs)
 {
-    AudioStreamPacketDescription* tmpPackages = (AudioStreamPacketDescription*)malloc(inNumberPacketDescriptions *sizeof(AudioStreamPacketDescription));
-    memcpy(tmpPackages,inPacketDescs,inNumberPacketDescriptions * sizeof(AudioStreamPacketDescription));
 
-    emit newWaveBuffer(pBuffer->mAudioData, pBuffer->mAudioDataByteSize, inNumberPacketDescriptions, tmpPackages);
+    emit newWaveBuffer(pBuffer->mAudioData, pBuffer->mAudioDataByteSize);
 
     qreal level = 0;
     UInt32 size;

src/podcast/quicktime/UBAudioQueueRecorder.h

@@ -57,23 +57,29 @@ class UBAudioQueueRecorder : public QObject
             return mLastErrorMessage;
         }
 
-        static AudioStreamBasicDescription audioFormat()
+        AudioStreamBasicDescription * audioFormat()
         {
-            return sAudioFormat;
+            return &sAudioFormat;
         }
 
     signals:
 
-        void newWaveBuffer(void* pBuffer, long pLength, int inNumberPacketDescriptions, const AudioStreamPacketDescription *inPacketDescs);
+        void newWaveBuffer(void* pBuffer, 
+                           long pLength);
 
         void audioLevelChanged(quint8 level);
 
     private:
-        static void audioQueueInputCallback (void *inUserData, AudioQueueRef inAQ,
-            AudioQueueBufferRef inBuffer, const AudioTimeStamp *inStartTime,
-            UInt32 inNumberPacketDescriptions, const AudioStreamPacketDescription *inPacketDescs);
-
-        void emitNewWaveBuffer(AudioQueueBufferRef pBuffer, int inNumberPacketDescriptions, const AudioStreamPacketDescription *inPacketDescs);
+        static void audioQueueInputCallback (void *inUserData, 
+                                             AudioQueueRef inAQ,
+                                             AudioQueueBufferRef inBuffer,
+                                             const AudioTimeStamp *inStartTime,
+                                             UInt32 inNumberPacketDescriptions, 
+                                             const AudioStreamPacketDescription *inPacketDescs);
+
+        void emitNewWaveBuffer(AudioQueueBufferRef pBuffer, 
+                               int inNumberPacketDescriptions,
+                               const AudioStreamPacketDescription *inPacketDescs);
 
         void emitAudioLevelChanged(quint8 level);
 

src/podcast/quicktime/UBQuickTimeFile.h

@@ -31,6 +31,7 @@
 #include <QtCore>
 
 #include <CoreVideo/CoreVideo.h>
+#include <CoreMedia/CoreMedia.h>
 
 #include "UBAudioQueueRecorder.h"
 
@@ -92,40 +93,50 @@ class UBQuickTimeFile : public QThread
     protected:
         void run();
 
+    private slots:
+        void appendAudioBuffer(void* pBuffer, long pLength);
 
     private:
 
         bool beginSession();
-
-        void appendVideoFrame(CVPixelBufferRef pixelBuffer, long msTimeStamp);
-
         void setLastErrorMessage(const QString& error);
 
+
+        void appendVideoFrame(CVPixelBufferRef pixelBuffer, long msTimeStamp);
         bool flushPendingFrames();
 
-        volatile CVPixelBufferPoolRef mCVPixelBufferPool;
         
+        volatile bool mShouldStopCompression;
+        volatile bool mCompressionSessionRunning;
+        volatile int mPendingFrames;
+
+        QString mSpatialQuality;
         
         int mFramesPerSecond;
         QSize mFrameSize;
         QString mVideoFileName;
-        long mTimeScale;
 
         bool mRecordAudio;
         
-        QString mLastErrorMessage;
         
-        QString mSpatialQuality;
+        AssetWriterPTR mVideoWriter;
 
-        volatile bool mShouldStopCompression;
-        volatile bool mCompressionSessionRunning;
+        AssetWriterInputPTR mVideoWriterInput;
+        AssetWriterInputAdaptorPTR mAdaptor;
+
+        AssetWriterInputPTR mAudioWriterInput;
+
+        QPointer<UBAudioQueueRecorder> mWaveRecorder;
+        CFAbsoluteTime mStartTime;
+
+        CMAudioFormatDescriptionRef mAudioFormatDescription;
+        
+        long mTimeScale;
+
+        QString mLastErrorMessage;
 
         QString mAudioRecordingDeviceName;
-        volatile int mPendingFrames;
 
-        AssetWriterPTR mVideoWriter;
-        AssetWriterInputPTR mVideoWriterInput;
-        AssetWriterInputAdaptorPTR mAdaptor;
 };
 
 #endif /* UBQUICKTIMEFILE_H_ */

src/podcast/quicktime/UBQuickTimeFile.mm

@@ -47,21 +47,20 @@ UBQuickTimeFile::UBQuickTimeFile(QObject * pParent)
     , mVideoWriter(0)
     , mVideoWriterInput(0)
     , mAdaptor(0)
-    , mCVPixelBufferPool(0)
     , mFramesPerSecond(-1)
-    , mTimeScale(100)
+    , mTimeScale(1000)
     , mRecordAudio(true)
     , mShouldStopCompression(false)
     , mCompressionSessionRunning(false)
     , mPendingFrames(0)
 {
-    // NOOP
 }
 
 
 UBQuickTimeFile::~UBQuickTimeFile()
 {
-    // NOOP
+    // destruction of mWaveRecorder is handled by endSession()
+
 }
 
 bool UBQuickTimeFile::init(const QString& pVideoFileName, const QString& pProfileData, int pFramesPerSecond
@@ -126,7 +125,11 @@ void UBQuickTimeFile::run()
 }
 
 /**
-  * \brief Initialize the AVAssetWriter, which handles writing the media to file
+  * \brief Begin the recording session; initialize the audio/video writer
+  * \return true if the session was initialized successfully
+  * 
+  * This function initializes the AVAssetWriter and associated video and audio inputs.
+  * Video is encoded as H264; audio is encoded as AAC.
   */
 bool UBQuickTimeFile::beginSession()
 {
@@ -147,6 +150,11 @@ bool UBQuickTimeFile::beginSession()
 
     mVideoWriter.movieTimeScale = mTimeScale;
     
+
+
+    // Video
+    //
+
     int frameWidth = mFrameSize.width();
     int frameHeight = mFrameSize.height();
 
@@ -175,18 +183,72 @@ bool UBQuickTimeFile::beginSession()
                         assetWriterInputPixelBufferAdaptorWithAssetWriterInput:mVideoWriterInput
                         sourcePixelBufferAttributes:pixelBufSettings] retain];
 
-    
-    
-    // Add the input(s) to the assetWriter
     NSCParameterAssert([mVideoWriter canAddInput:mVideoWriterInput]);
     [mVideoWriter addInput:mVideoWriterInput];
 
 
-    // begin the writing session
+
+    // Audio
+    //
+
+    if(mRecordAudio) {
+        mWaveRecorder = new UBAudioQueueRecorder();
+        
+        // Get the audio format description from mWaveRecorder
+        CMAudioFormatDescriptionCreate(kCFAllocatorDefault, mWaveRecorder->audioFormat(),
+                                       0, NULL, 0, NULL, NULL,
+                                       &mAudioFormatDescription);
+
+        if(mWaveRecorder->init(mAudioRecordingDeviceName)) {
+            connect(mWaveRecorder, &UBAudioQueueRecorder::newWaveBuffer,
+                    this, &UBQuickTimeFile::appendAudioBuffer);
+
+            connect(mWaveRecorder, SIGNAL(audioLevelChanged(quint8)),
+                    this, SIGNAL(audioLevelChanged(quint8)));
+        }
+        else {
+            setLastErrorMessage(mWaveRecorder->lastErrorMessage());
+            mWaveRecorder->deleteLater();
+            mRecordAudio = false;
+            break;
+        }
+
+        // Audio is mono, and compressed to AAC at 128kbps
+    
+        AudioChannelLayout audioChannelLayout = {
+            .mChannelLayoutTag = kAudioChannelLayoutTag_Mono,
+            .mChannelBitmap = 0,
+            .mNumberChannelDescriptions = 0
+        };
+
+        NSData *channelLayoutAsData = [NSData dataWithBytes:&audioChannelLayout 
+                                              length:offsetof(AudioChannelLayout, mChannelDescriptions)];
+        
+        NSDictionary * compressionAudioSettings = @{
+            AVFormatIDKey         : [NSNumber numberWithUnsignedInt:kAudioFormatMPEG4AAC],
+            AVEncoderBitRateKey   : [NSNumber numberWithInteger:128000],
+            AVSampleRateKey       : [NSNumber numberWithInteger:44100],
+            AVChannelLayoutKey    : channelLayoutAsData,
+            AVNumberOfChannelsKey : [NSNumber numberWithUnsignedInteger:1]
+        };
+
+        mAudioWriterInput = [[AVAssetWriterInput assetWriterInputWithMediaType:AVMediaTypeAudio 
+                                               outputSettings:compressionAudioSettings] retain];
+        
+        NSCParameterAssert([mVideoWriter canAddInput:mAudioWriterInput]);
+        [mVideoWriter addInput:mAudioWriterInput];
+        
+        qDebug() << "audio writer input created and added";
+    }
+
+
+    // Begin the writing session
     bool canStartWriting = [mVideoWriter startWriting];
     [mVideoWriter startSessionAtSourceTime:CMTimeMake(0, mTimeScale)];
 
-    // return true if everything was created and started successfully
+    mStartTime = CFAbsoluteTimeGetCurrent(); // used for audio timestamp calculation
+
+    
     return (mVideoWriter != nil) && (mVideoWriterInput != nil) && canStartWriting;
 }
 
@@ -201,10 +263,18 @@ void UBQuickTimeFile::endSession()
     [mAdaptor release];
     [mVideoWriterInput release];
     [mVideoWriter release];
+    [mAudioWriterInput release];
 
     mAdaptor = nil;
     mVideoWriterInput = nil;
     mVideoWriter = nil;
+    mAudioWriterInput = nil;
+
+    if (mWaveRecorder) {
+        mWaveRecorder->close();
+        mWaveRecorder->deleteLater();
+    }
+
 }
 
 /**
@@ -235,11 +305,11 @@ CVPixelBufferRef UBQuickTimeFile::newPixelBuffer()
 
 /**
  * \brief Add a frame to the pixel buffer adaptor
+ * \param pixelBuffer The CVPixelBufferRef (video frame) to add to the movie
+ * \param msTimeStamp Timestamp, in milliseconds, of the frame
  */
 void UBQuickTimeFile::appendVideoFrame(CVPixelBufferRef pixelBuffer, long msTimeStamp)
 {
-    //qDebug() << "adding video frame at time: " << msTimeStamp;
-
     CMTime t = CMTimeMake((msTimeStamp * mTimeScale / 1000.0), mTimeScale); 
 
     bool added = [mAdaptor appendPixelBuffer: pixelBuffer
@@ -252,11 +322,79 @@ void UBQuickTimeFile::appendVideoFrame(CVPixelBufferRef pixelBuffer, long msTime
     CVPixelBufferRelease(pixelBuffer);
 }
 
+
+
+
+/**
+ * \brief Append an AudioQueue Buffer to the audio AVAssetWriterInput
+ * \param pBuffer The AudioQueueBufferRef to add. Must be uncompressed (LPCM).
+ * \param pLength The length of the buffer, in Bytes
+ *
+ * This function serves as an interface between the low-level audio stream
+ * (implemented in the UBAudioQueueRecorder class) and the recording, handled
+ * by the AVAssetWriterInput instance mAudioWriterInput.
+ */
+void UBQuickTimeFile::appendAudioBuffer(void* pBuffer, 
+                                        long pLength)
+{
+    if(!mRecordAudio) 
+        return;
+
+
+    // CMSampleBuffers require a CMBlockBuffer to hold the media data; we
+    // create a blockBuffer here from the AudioQueueBuffer's data.
+    CMBlockBufferRef blockBuffer;
+    CMBlockBufferCreateWithMemoryBlock(kCFAllocatorDefault,
+                                       pBuffer,
+                                       pLength,
+                                       kCFAllocatorNull,
+                                       NULL,
+                                       0,
+                                       pLength,
+                                       kCMBlockBufferAssureMemoryNowFlag,
+                                       &blockBuffer);
+
+
+    // Timestamp of current sample
+    CFAbsoluteTime currentTime = CFAbsoluteTimeGetCurrent();
+    CFTimeInterval elapsedTime = currentTime - mStartTime;
+    CMTime timeStamp = CMTimeMake(elapsedTime * mTimeScale, mTimeScale);
+
+    // Number of samples in the buffer
+    long nSamples = pLength / mWaveRecorder->audioFormat()->mBytesPerFrame;
+
+    CMSampleBufferRef sampleBuffer;
+    CMAudioSampleBufferCreateWithPacketDescriptions(kCFAllocatorDefault,
+                                                    blockBuffer,
+                                                    true,
+                                                    NULL,
+                                                    NULL,
+                                                    mAudioFormatDescription, 
+                                                    nSamples,
+                                                    timeStamp, 
+                                                    NULL,
+                                                    &sampleBuffer);
+
+
+    // Add the audio sample to the asset writer input
+    if ([mAudioWriterInput isReadyForMoreMediaData])
+        if(![mAudioWriterInput appendSampleBuffer:sampleBuffer])
+            setLastErrorMessage(QString("Failed to append sample buffer to audio input: %1").arg(err));
+    
+    else 
+        setLastErrorMessage(QString("Audio Writer not ready; sample dropped"));
+    
+
+    CFRelease(sampleBuffer);
+    CFRelease(blockBuffer);
+    // The audioQueueBuffers are all freed when UBAudioQueueRecorder::close() is called
+}
+
+/**
+ * \brief Print an error message to the terminal, and store it
+ */
 void UBQuickTimeFile::setLastErrorMessage(const QString& error)
 {
     mLastErrorMessage = error;
     qWarning() << "UBQuickTimeFile error" << error;
 }
-
-
-