Using AVFoundation with AVCaptureAudioDataOutput for Audio playback

  • Hello

    I am trying to write an external plugin that simply reads audio from the built in microphone, obtains the raw audio data, and passes it to an SDK API.
    My setup is very simple, I have a single session, with a single input device (built in microphone) and a single output (AVCaptureAudioDataOutput)
    I Initialize the Session, Input, and Output as follow:

                // Obtain the device and retain it

                _pAudioCaptureInput = [pAVInputDevice retain];

                // Initialize Capture Session

                _dispatchQueue = dispatch_get_main_queue();

                _pCaptureSession = [[AVCaptureSession alloc] init];

                _pCaptureSession.sessionPreset = AVCaptureSessionPresetHigh;

                // Put the Device inside an Input

                pInputDevice = [[AVCaptureDeviceInput alloc] initWithDevice:_pAudioCaptureInput error:&pError];

                [_pCaptureSession addInput:pInputDevice]; // Add Audio Input to Session

                [pInputDevice release];

                pInputDevice = nil;

                // Setup the Output Configurations

                NSDictionary *pAudioSettings = [NSDictionary dictionaryWithObjectsAndKeys:

                                                      [NSNumber numberWithInt:kAudioFormatLinearPCM], AVFormatIDKey,

                                                      [NSNumber numberWithInt:44100], AVSampleRateKey,

                                                      [NSNumber numberWithInt:2], AVNumberOfChannelsKey,

                                                      [NSNumber numberWithInt:16], AVLinearPCMBitDepthKey,

                                                      [NSNumber numberWithBool:NO], AVLinearPCMIsFloatKey,

                                                      [NSNumber numberWithBool:NO], AVLinearPCMIsBigEndianKey,


                // Create the Output

                _pAudioCaptureOutput = [[AVCaptureAudioDataOutput alloc] init];

                [_pAudioCaptureOutput setAudioSettings:pAudioSettings]; // Apply Output Settings

                [_pCaptureSession addOutput:_pAudioCaptureOutput]; // Add Output to Session

                // Create the Output Delegate that will receive the SampleBuffer callbacks

                _pAudioCaptureDelegate = [[ACaptureDelegate alloc] init];

                [_pAudioCaptureOutput setSampleBufferDelegate:_pAudioCaptureDelegate


                // Remember the Device Name

                _deviceName = [_pAudioCaptureInput.localizedName UTF8String];

    I start the session running by calling startRunning on the session.  My callback delegate is notified with Audio Samples.  The follow is the callback implementation:

    - (void)captureOutput:(AVCaptureOutput *)captureOutput didOutputSampleBuffer:(CMSampleBufferRef)sampleBuffer fromConnection:(AVCaptureConnection *)connection


    // Save the Timescale

        CMTime timestamp = CMSampleBufferGetOutputPresentationTimeStamp(sampleBuffer);

        if(_timescale == -1) {

            _timescale = timestamp.timescale;


        CMItemCount numSamples = CMSampleBufferGetNumSamples(sampleBuffer);

        // Get Audio Data

        CMBlockBufferRef dataBuffer = CMSampleBufferGetDataBuffer(sampleBuffer);

        zuint32 dataLength = CMBlockBufferGetDataLength(dataBuffer);

        AVCaptureAudioDataOutput *pOutput = static_cast<AVCaptureAudioDataOutput *>(captureOutput);

        NSDictionary *pAudioSettings = pOutput.audioSettings;

        // Retrieve Audio Settings Information from AVOutputDevice

        NSNumber *audioSettingValue = nil;

        zint32 sampleRate = 44100;

        audioSettingValue = [pAudioSettings objectForKey:AVSampleRateKey];

        if(audioSettingValue) {

            sampleRate = [audioSettingValue intValue];


        zint32 bitDepth = 16;

        audioSettingValue = [pAudioSettings objectForKey:AVLinearPCMBitDepthKey];

        if(audioSettingValue) {

            bitDepth = [audioSettingValue intValue];


        zint32 channels = 2;

        audioSettingValue = [pAudioSettings objectForKey:AVNumberOfChannelsKey];

        if(audioSettingValue) {

            channels = [audioSettingValue intValue];


        zuint8 *pAudioData = new zuint8[dataLength];

        OSStatus err = CMBlockBufferCopyDataBytes(dataBuffer, 0, dataLength, pAudioData);

        if(err == noErr) {

    // Populate SDK Audio structure

            _audioBuffer.format = kFormatPCM;

            _audioBuffer.dataLength = dataLength;

            _audioBuffer.timeStamp = 0;

            _audioBuffer.flags = kBufferFlagTimeValid;

            _audioBuffer.sampleRate = sampleRate;

            _audioBuffer.bitsPerSample = bitDepth;

            _audioBuffer.channels = channels;

   = pAudioData; // The Raw Audio Data

    // Send Audio Data to SDK API

            _pAudioDataArriveCallback(_pAudioDataArriveContext, &_audioBuffer);

            delete [] pAudioData;

            pAudioData = NULL;



    The Problem is, the audio is playing back with a higher pitch.  Like someone is fast forwarding.  If I adjust the SampleRate in the Output Audio Settings, it affects the play out.
    Higher I set it, the less squeaky it becomes, but never correct.  Is there any initialization step I am missing?

    I have not been able to play the raw audio data back myself without the main app as I am new to AVFoundation and CoreAudio.
    If someone has sample code of getting this up, I would much appreciate it.

    -San Saeteurn
  • You should ask about audio questions on the coreaudio-dev list, not here. I don’t think the audio experts read this list. (It’s only recently that there have been real Cocoa-level APIs for audio at all.)

previous month may 2013 next month
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    
Go to today