Play audio on iOS from a memory data stream - ios

Question

I am porting an audio library to iOS allowing to play audio streams fed from callbacks. The user provides a callback returning raw PCM data, and I need to have this data be played. Moreover, the library must be able to play multiple streams at once.
I figured I would need to use AVFoundation, but it seems like AVAudioPlayer does not support streamed audio buffers, and all the streaming documentation I could find used data coming directly from the network. What is the API I should use here?
Thanks in advance!
By the way, I am not using the Apple libraries through Swift or Objective-C. However I assume everything is exposed still, so an example in Swift would be greatly appreciated anyway!

Answer 1

You need to initialise:
The Audio Session to use input audio unit and output.
-(SInt32) audioSessionInitialization:(SInt32)preferred_sample_rate {
// - - - - - - Audio Session initialization
NSError *audioSessionError = nil;
session = [AVAudioSession sharedInstance];
// disable AVAudioSession
[session setActive:NO error:&audioSessionError];
// set category - (PlayAndRecord to use input and output session AudioUnits)
[session setCategory:AVAudioSessionCategoryPlayAndRecord withOptions:AVAudioSessionCategoryOptionDefaultToSpeaker error:&audioSessionError];
double preferredSampleRate = 441000;
[session setPreferredSampleRate:preferredSampleRate error:&audioSessionError];
// enable AVAudioSession
[session setActive:YES error:&audioSessionError];
// Configure notification for device output change (speakers/headphones)
[[NSNotificationCenter defaultCenter] addObserver:self
selector:#selector(routeChange:)
name:AVAudioSessionRouteChangeNotification
object:nil];
// - - - - - - Create audio engine
[self audioEngineInitialization];
return [session sampleRate];
}
The Audio Engine
-(void) audioEngineInitialization{
engine = [[AVAudioEngine alloc] init];
inputNode = [engine inputNode];
outputNode = [engine outputNode];
[engine connect:inputNode to:outputNode format:[inputNode inputFormatForBus:0]];
AudioStreamBasicDescription asbd_player;
asbd_player.mSampleRate = session.sampleRate;
asbd_player.mFormatID = kAudioFormatLinearPCM;
asbd_player.mFormatFlags = kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked;
asbd_player.mFramesPerPacket = 1;
asbd_player.mChannelsPerFrame = 2;
asbd_player.mBitsPerChannel = 16;
asbd_player.mBytesPerPacket = 4;
asbd_player.mBytesPerFrame = 4;
OSStatus status;
status = AudioUnitSetProperty(inputNode.audioUnit,
kAudioUnitProperty_StreamFormat,
kAudioUnitScope_Input,
0,
&asbd_player,
sizeof(asbd_player));
// Add the render callback for the ioUnit: for playing
AURenderCallbackStruct callbackStruct;
callbackStruct.inputProc = engineInputCallback; ///CALLBACK///
callbackStruct.inputProcRefCon = (__bridge void *)(self);
status = AudioUnitSetProperty(inputNode.audioUnit,
kAudioUnitProperty_SetRenderCallback,
kAudioUnitScope_Input,//Global
kOutputBus,
&callbackStruct,
sizeof(callbackStruct));
[engine prepare];
}
The Audio Engine callback
static OSStatus engineInputCallback(void *inRefCon,
AudioUnitRenderActionFlags *ioActionFlags,
const AudioTimeStamp *inTimeStamp,
UInt32 inBusNumber,
UInt32 inNumberFrames,
AudioBufferList *ioData)
{
// the reference to the audio controller where you get the stream data
MyAudioController *ac = (__bridge MyAudioController *)(inRefCon);
// in practice we will only ever have 1 buffer, since audio format is mono
for (int i = 0; i < ioData->mNumberBuffers; i++) {
AudioBuffer buffer = ioData->mBuffers[i];
// copy stream buffer data to output buffer
UInt32 size = min(buffer.mDataByteSize, ac.playbackBuffer.mDataByteSize);
memcpy(buffer.mData, ac.streamBuffer.mData, size);
buffer.mDataByteSize = size; // indicate how much data we wrote in the buffer
}
return noErr;
}