Writing buffer of audio samples to aac file using ExtAudioFileWrite for iOS

Question

UPDATE: I have figured this out and posted my solution as an answer to my own question (below)

I am trying to write a simple buffer of audio samples to a file using ExtAudioFileWrite in AAC format.

I have achieved this with the code below to write a mono buffer to a .wav file - however, I cannot do this for stereo or for AAC files which is what I want to do.

Here is what I have so far...

CFStringRef fPath;
fPath = CFStringCreateWithCString(kCFAllocatorDefault,
                                  "/path/to/my/audiofile/audiofile.wav",
                                  kCFStringEncodingMacRoman);


OSStatus err;

int mChannels = 1;
UInt32 totalFramesInFile = 100000;

Float32 *outputBuffer = (Float32 *)malloc(sizeof(Float32) * (totalFramesInFile*mChannels)); 


////////////// Set up Audio Buffer List ////////////

AudioBufferList outputData; 
outputData.mNumberBuffers = 1;
outputData.mBuffers[0].mNumberChannels = mChannels; 
outputData.mBuffers[0].mDataByteSize = 4 * totalFramesInFile * mChannels;
outputData.mBuffers[0].mData = outputBuffer;

Float32 audioFile[totalFramesInFile*mChannels];


for (int i = 0;i < totalFramesInFile*mChannels;i++)
{
    audioFile[i] = ((Float32)(rand() % 100))/100.0;
    audioFile[i] = audioFile[i]*0.2;
}

outputData.mBuffers[0].mData = &audioFile;


CFURLRef fileURL = CFURLCreateWithFileSystemPath(kCFAllocatorDefault,fPath,kCFURLPOSIXPathStyle,false);

ExtAudioFileRef audiofileRef;

// WAVE FILES

AudioFileTypeID fileType = kAudioFileWAVEType;
AudioStreamBasicDescription clientFormat;
clientFormat.mSampleRate = 44100.0;
clientFormat.mFormatID = kAudioFormatLinearPCM;
clientFormat.mFormatFlags = 12;
clientFormat.mBitsPerChannel = 16;
clientFormat.mChannelsPerFrame = mChannels;
clientFormat.mBytesPerFrame = 2*clientFormat.mChannelsPerFrame;
clientFormat.mFramesPerPacket = 1;
clientFormat.mBytesPerPacket = 2*clientFormat.mChannelsPerFrame;


// open the file for writing
err = ExtAudioFileCreateWithURL((CFURLRef)fileURL, fileType, &clientFormat, NULL, kAudioFileFlags_EraseFile, &audiofileRef);

if (err != noErr)
{
    cout << "Problem when creating audio file: " << err << "\n";
}

// tell the ExtAudioFile API what format we'll be sending samples in
err = ExtAudioFileSetProperty(audiofileRef, kExtAudioFileProperty_ClientDataFormat, sizeof(clientFormat), &clientFormat);

if (err != noErr)
{
    cout << "Problem setting audio format: " << err << "\n";
}

UInt32 rFrames = (UInt32)totalFramesInFile;
// write the data
err = ExtAudioFileWrite(audiofileRef, rFrames, &outputData);

if (err != noErr)
{
    cout << "Problem writing audio file: " << err << "\n";
}

// close the file
ExtAudioFileDispose(audiofileRef);



NSLog(@"Done!");

My specific questions are:

• How do I set up the AudioStreamBasicDescription for AAC?
• Why can't I get stereo to work properly here? If I set the number of channels ('mChannels') to 2 then I get the left channel correctly and distortion in the right channel.

I'd very much appreciate any help - I think I've read almost every page I can find on this and am none the wiser as, while there are similar questions, they usually derive the AudioStreamBasicDescription parameters from some input audio file, which I cannot see the result of. The Apple documentation is no help either.

Many thanks in advance,

Adam

Answer 1

Ok, after some exploration I have figured it out. I have wrapped it as a function that writes random noise to a file. Specifically, it can:

• write .wav or .m4a files
• write mono or stereo in either format
• write the file to a specified path

The function arguments are:

• path to audio file to be created
• number of channels (max 2)
• boolean: compress with m4a (if false, use pcm)

For a stereo M4A file, the function should be called as:

writeNoiseToAudioFile("/path/to/my/audiofile.m4a",2,true);

The source of the function follows. I have tried to comment it as much as possible - I hope it is correct, it certainly works for me, but please say "Adam, you've done this a bit wrong" if there is something I've missed. Good luck! Here is the code:

void writeNoiseToAudioFile(char *fName,int mChannels,bool compress_with_m4a)
{
OSStatus err; // to record errors from ExtAudioFile API functions

// create file path as CStringRef
CFStringRef fPath;
fPath = CFStringCreateWithCString(kCFAllocatorDefault,
                                  fName,
                                  kCFStringEncodingMacRoman);


// specify total number of samples per channel
UInt32 totalFramesInFile = 100000;      

/////////////////////////////////////////////////////////////////////////////
////////////// Set up Audio Buffer List For Interleaved Audio ///////////////
/////////////////////////////////////////////////////////////////////////////

AudioBufferList outputData; 
outputData.mNumberBuffers = 1;
outputData.mBuffers[0].mNumberChannels = mChannels;    
outputData.mBuffers[0].mDataByteSize = sizeof(AudioUnitSampleType)*totalFramesInFile*mChannels;



/////////////////////////////////////////////////////////////////////////////
//////// Synthesise Noise and Put It In The AudioBufferList /////////////////
/////////////////////////////////////////////////////////////////////////////

// create an array to hold our audio
AudioUnitSampleType audioFile[totalFramesInFile*mChannels];

// fill the array with random numbers (white noise)
for (int i = 0;i < totalFramesInFile*mChannels;i++)
{
    audioFile[i] = ((AudioUnitSampleType)(rand() % 100))/100.0;
    audioFile[i] = audioFile[i]*0.2;
    // (yes, I know this noise has a DC offset, bad)
}

// set the AudioBuffer to point to the array containing the noise
outputData.mBuffers[0].mData = &audioFile;


/////////////////////////////////////////////////////////////////////////////
////////////////// Specify The Output Audio File Format /////////////////////
/////////////////////////////////////////////////////////////////////////////


// the client format will describe the output audio file
AudioStreamBasicDescription clientFormat;

// the file type identifier tells the ExtAudioFile API what kind of file we want created
AudioFileTypeID fileType;

// if compress_with_m4a is tru then set up for m4a file format
if (compress_with_m4a)
{
    // the file type identifier tells the ExtAudioFile API what kind of file we want created
    // this creates a m4a file type
    fileType = kAudioFileM4AType;

    // Here we specify the M4A format
    clientFormat.mSampleRate         = 44100.0;
    clientFormat.mFormatID           = kAudioFormatMPEG4AAC;
    clientFormat.mFormatFlags        = kMPEG4Object_AAC_Main;
    clientFormat.mChannelsPerFrame   = mChannels;
    clientFormat.mBytesPerPacket     = 0;
    clientFormat.mBytesPerFrame      = 0;
    clientFormat.mFramesPerPacket    = 1024;
    clientFormat.mBitsPerChannel     = 0;
    clientFormat.mReserved           = 0;
}
else // else encode as PCM
{
    // this creates a wav file type
    fileType = kAudioFileWAVEType;

    // This function audiomatically generates the audio format according to certain arguments
    FillOutASBDForLPCM(clientFormat,44100.0,mChannels,32,32,true,false,false);
}



/////////////////////////////////////////////////////////////////////////////
///////////////// Specify The Format of Our Audio Samples ///////////////////
/////////////////////////////////////////////////////////////////////////////

// the local format describes the format the samples we will give to the ExtAudioFile API
AudioStreamBasicDescription localFormat;
FillOutASBDForLPCM (localFormat,44100.0,mChannels,32,32,true,false,false);



/////////////////////////////////////////////////////////////////////////////
///////////////// Create the Audio File and Open It /////////////////////////
/////////////////////////////////////////////////////////////////////////////

// create the audio file reference
ExtAudioFileRef audiofileRef;

// create a fileURL from our path
CFURLRef fileURL = CFURLCreateWithFileSystemPath(kCFAllocatorDefault,fPath,kCFURLPOSIXPathStyle,false);

// open the file for writing
err = ExtAudioFileCreateWithURL((CFURLRef)fileURL, fileType, &clientFormat, NULL, kAudioFileFlags_EraseFile, &audiofileRef);

if (err != noErr)
{
    cout << "Problem when creating audio file: " << err << "\n";
}


/////////////////////////////////////////////////////////////////////////////
///// Tell the ExtAudioFile API what format we'll be sending samples in /////
/////////////////////////////////////////////////////////////////////////////

// Tell the ExtAudioFile API what format we'll be sending samples in 
err = ExtAudioFileSetProperty(audiofileRef, kExtAudioFileProperty_ClientDataFormat, sizeof(localFormat), &localFormat);

if (err != noErr)
{
    cout << "Problem setting audio format: " << err << "\n";
}

/////////////////////////////////////////////////////////////////////////////
///////// Write the Contents of the AudioBufferList to the AudioFile ////////
/////////////////////////////////////////////////////////////////////////////

UInt32 rFrames = (UInt32)totalFramesInFile;
// write the data
err = ExtAudioFileWrite(audiofileRef, rFrames, &outputData);

if (err != noErr)
{
    cout << "Problem writing audio file: " << err << "\n";
}


/////////////////////////////////////////////////////////////////////////////
////////////// Close the Audio File and Get Rid Of The Reference ////////////
/////////////////////////////////////////////////////////////////////////////

// close the file
ExtAudioFileDispose(audiofileRef);


NSLog(@"Done!");
}

Don't forget to import the AudioToolbox Framework and to include the header file:

#import <AudioToolbox/AudioToolbox.h>