How to play live audio on iOS? - ios

Question

I have an IPCamera that requires the use of a custom library for connecting and communication. I have the video all taken care of, but I also want to give the user the option to listen to the audio that is recorded by the camera.
I receive the audio in the form of a byte stream (the audio is PCM u-law).
Since I don't read the data from a file or have an URL I can connect to, I think I would have to use something like AudioUnits or openAL to play my audio.
I tried to implement it with AudioUnits based on the examples I found online and this is what I have so far:
-(void) audioThread
{
char buffer[1024];
int size = 0;
boolean audioConfigured = false;
AudioComponentInstance audioUnit;
while (running) {
getAudioData(buffer,size); //fill buffer with my audio
int16_t* tempChar = (int16_t *)calloc(ret, sizeof(int16_t));
for (int i = 0; i < ret; i++) {
tempChar[i] = MuLaw_Decode(buf[i]);
}
uint8_t *data = NULL;
data = malloc(size);
data = memcpy(data, &tempChar, size);
CMBlockBufferRef blockBuffer = NULL;
OSStatus status = CMBlockBufferCreateWithMemoryBlock(NULL, data,
size,
kCFAllocatorNull, NULL,
0,
size,
0, &blockBuffer);
CMSampleBufferRef sampleBuffer = NULL;
// now I create my samplebuffer from the block buffer
if(status == noErr)
{
const size_t sampleSize = size;
status = CMSampleBufferCreate(kCFAllocatorDefault,
blockBuffer, true, NULL, NULL,
formatDesc, 1, 0, NULL, 1,
&sampleSize, &sampleBuffer);
}
AudioStreamBasicDescription audioBasic;
audioBasic.mBitsPerChannel = 16;
audioBasic.mBytesPerPacket = 2;
audioBasic.mBytesPerFrame = 2;
audioBasic.mChannelsPerFrame = 1;
audioBasic.mFormatFlags = kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked;
audioBasic.mFormatID = kAudioFormatLinearPCM;
audioBasic.mFramesPerPacket = 1;
audioBasic.mSampleRate = 48000;
audioBasic.mReserved = 0;
if(!audioConfigured)
{
//initialize the circular buffer
if(instance.decodingBuffer == NULL)
instance.decodingBuffer = malloc(sizeof(TPCircularBuffer));
if(!TPCircularBufferInit(instance.decodingBuffer, 1024))
continue;
AudioComponentDescription componentDescription;
componentDescription.componentType = kAudioUnitType_Output;
componentDescription.componentSubType = kAudioUnitSubType_RemoteIO;
componentDescription.componentManufacturer = kAudioUnitManufacturer_Apple;
componentDescription.componentFlags = 0;
componentDescription.componentFlagsMask = 0;
AudioComponent component = AudioComponentFindNext(NULL, &componentDescription);
if(AudioComponentInstanceNew(component, &audioUnit) != noErr) {
NSLog(#"Failed to initialize the AudioComponent");
continue;
}
//enable IO for playback
UInt32 flag = 1;
if(AudioUnitSetProperty(audioUnit, kAudioOutputUnitProperty_EnableIO, kAudioUnitScope_Output, 0, &flag, sizeof(flag)) != noErr) {
NSLog(#"Failed to enable IO for playback");
continue;
}
// set the format for the outputstream
if(AudioUnitSetProperty(audioUnit, kAudioUnitProperty_StreamFormat,
kAudioUnitScope_Output, 1, &audioBasic, sizeof(audioBasic)) != noErr) {
NSLog(#"Failed to set the format for the outputstream");
continue;
}
// set output callback
AURenderCallbackStruct callbackStruct;
callbackStruct.inputProc = playbackCallback;
callbackStruct.inputProcRefCon = (__bridge void*) self;
if(AudioUnitSetProperty(audioUnit, kAudioUnitProperty_SetRenderCallback, kAudioUnitScope_Global, 0, &callbackStruct, sizeof(callbackStruct))!= noErr) {
NSLog(#"Failed to Set output callback");
continue;
}
// Disable buffer allocation for the recorder (optional - do this if we want to pass in our own)
flag = 0;
status = AudioUnitSetProperty(audioUnit, kAudioUnitProperty_ShouldAllocateBuffer, kAudioUnitScope_Output, 1, &flag, sizeof(flag));
if(AudioUnitInitialize(audioUnit) != noErr) {
NSLog(#"Failed to initialize audioUnits");
}
if(AudioOutputUnitStart(audioUnit)!= noErr) {
NSLog(#"[thread_ReceiveAudio] Failed to start audio");
}
audioConfigured = true;
}
AudioBufferList bufferList ;
if (sampleBuffer!=NULL) {
CMSampleBufferGetAudioBufferListWithRetainedBlockBuffer(sampleBuffer, NULL, &bufferList, sizeof(bufferList), NULL, NULL, kCMSampleBufferFlag_AudioBufferList_Assure16ByteAlignment, &blockBuffer);
UInt64 size = CMSampleBufferGetTotalSampleSize(sampleBuffer);
// Put audio into circular buffer
TPCircularBufferProduceBytes(self.decodingBuffer, bufferList.mBuffers[0].mData, size);
//TPCircularBufferCopyAudioBufferList(self.decodingBuffer, &bufferList, NULL, kTPCircularBufferCopyAll, NULL);
CFRelease(sampleBuffer);
CFRelease(blockBuffer);
}
}
//stop playing audio
if(audioConfigured){
if(AudioOutputUnitStop(audioUnit)!= noErr) {
NSLog(#"[thread_ReceiveAudio] Failed to stop audio");
}
else{
//clean up audio
AudioComponentInstanceDispose(audioUnit);
}
}
}
int16_t MuLaw_Decode(int8_t number)
{
const uint16_t MULAW_BIAS = 33;
uint8_t sign = 0, position = 0;
int16_t decoded = 0;
number = ~number;
if (number & 0x80)
{
number &= ~(1 << 7);
sign = -1;
}
position = ((number & 0xF0) >> 4) + 5;
decoded = ((1 << position) | ((number & 0x0F) << (position - 4))
| (1 << (position - 5))) - MULAW_BIAS;
return (sign == 0) ? (decoded) : (-(decoded));
}
static OSStatus playbackCallback(void *inRefCon,
AudioUnitRenderActionFlags *ioActionFlags,
const AudioTimeStamp *inTimeStamp,
UInt32 inBusNumber,
UInt32 inNumberFrames,
AudioBufferList *ioData) {
int bytesToCopy = ioData->mBuffers[0].mDataByteSize;
SInt16 *targetBuffer = (SInt16*)ioData->mBuffers[0].mData;
int32_t availableBytes;
SInt16 *buffer = TPCircularBufferTail(instance.decodingBuffer, &availableBytes);
int sampleCount = MIN(bytesToCopy, availableBytes);
memcpy(targetBuffer, buffer, MIN(bytesToCopy, availableBytes));
TPCircularBufferConsume(self.decodingBuffer, sampleCount);
return noErr;
}
The code above doesn't produce any errors, but won't play any sound. I though I could set the audio through the bufferList in the recordCallback, but it is never called.
So my question is: How do I play audio from a byte stream on iOS?

Answer 1

I decided to look at the project with fresh eyes. I got rid of most of the code and got it to work now. It is not pretty, but at least it runs for now. For example: I had to set my sample rate to 4000, otherwise it would play to fast and I still have performance issues. Anyway this is what I came up with:
#define BUFFER_SIZE 1024
#define NUM_CHANNELS 2
#define kOutputBus 0
#define kInputBus 1
-(void) main
{
char buf[BUFFER_SIZE];
int size;
runloop: while (self.running) {
getAudioData(&buf, size);
if(!self.configured) {
if(![self activateAudioSession])
continue;
self.configured = true;
}
TPCircularBufferProduceBytes(self.decodingBuffer, buf, size);
}
//stop audiounits
AudioOutputUnitStop(self.audioUnit);
AudioComponentInstanceDispose(self.audioUnit);
if (self.decodingBuffer != NULL) {
TPCircularBufferCleanup(self.decodingBuffer);
}
}
static void audioSessionInterruptionCallback(void *inUserData, UInt32 interruptionState) {
if (interruptionState == kAudioSessionEndInterruption) {
AudioSessionSetActive(YES);
AudioOutputUnitStart(self.audioUnit);
}
if (interruptionState == kAudioSessionBeginInterruption) {
AudioOutputUnitStop(self.audioUnit);
}
}
static OSStatus playbackCallback(void *inRefCon,
AudioUnitRenderActionFlags *ioActionFlags,
const AudioTimeStamp *inTimeStamp,
UInt32 inBusNumber,
UInt32 inNumberFrames,
AudioBufferList *ioData) {
// Notes: ioData contains buffers (may be more than one!)
// Fill them up as much as you can. Remember to set the size value in each buffer to match how much data is in the buffer.
if (!self.running ) {
return -1;
}
int bytesToCopy = ioData->mBuffers[0].mDataByteSize;
SInt16 *targetBuffer = (SInt16*)ioData->mBuffers[0].mData;
// Pull audio from playthrough buffer
int32_t availableBytes;
if(self.decodingBuffer == NULL || self.decodingBuffer->length < 1) {
NSLog(#"buffer is empty");
return 0;
}
SInt16 *buffer = TPCircularBufferTail(self.decodingBuffer, &availableBytes);
int sampleCount = MIN(bytesToCopy, availableBytes);
memcpy(targetBuffer, buffer, sampleCount);
TPCircularBufferConsume(self.decodingBuffer, sampleCount);
return noErr;
}
- (BOOL) activateAudioSession {
if (!self.activated_) {
OSStatus result;
result = AudioSessionInitialize(NULL,
NULL,
audioSessionInterruptionCallback,
(__bridge void *)(self));
if (kAudioSessionAlreadyInitialized != result)
[self checkError:result message:#"Couldn't initialize audio session"];
[self setupAudio]
self.activated_ = YES;
}
return self.activated_;
}
- (void) setupAudio
{
OSStatus status;
// Describe audio component
AudioComponentDescription desc;
desc.componentType = kAudioUnitType_Output;
desc.componentSubType = kAudioUnitSubType_RemoteIO;
desc.componentFlags = 0;
desc.componentFlagsMask = 0;
desc.componentManufacturer = kAudioUnitManufacturer_Apple;
// Get component
AudioComponent inputComponent = AudioComponentFindNext(NULL, &desc);
// Get audio units
AudioComponentInstanceNew(inputComponent, &_audioUnit);
// // Enable IO for recording
// UInt32 flag = 1;
// status = AudioUnitSetProperty(audioUnit,
// kAudioOutputUnitProperty_EnableIO,
// kAudioUnitScope_Input,
// kInputBus,
// &flag,
// sizeof(flag));
// Enable IO for playback
UInt32 flag = 1;
AudioUnitSetProperty(_audioUnit,
kAudioOutputUnitProperty_EnableIO,
kAudioUnitScope_Output,
kOutputBus,
&flag,
sizeof(flag));
// Describe format
AudioStreamBasicDescription format;
format.mSampleRate = 4000;
format.mFormatID = kAudioFormatULaw; //kAudioFormatULaw
format.mFormatFlags = kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked;//
format.mBitsPerChannel = 8 * sizeof(char);
format.mChannelsPerFrame = NUM_CHANNELS;
format.mBytesPerFrame = sizeof(char) * NUM_CHANNELS;
format.mFramesPerPacket = 1;
format.mBytesPerPacket = format.mBytesPerFrame * format.mFramesPerPacket;
format.mReserved = 0;
self.audioFormat = format;
// Apply format
AudioUnitSetProperty(_audioUnit,
kAudioUnitProperty_StreamFormat,
kAudioUnitScope_Output,
kInputBus,
&_audioFormat,
sizeof(_audioFormat));
AudioUnitSetProperty(_audioUnit,
kAudioUnitProperty_StreamFormat,
kAudioUnitScope_Input,
kOutputBus,
&_audioFormat,
sizeof(_audioFormat));
// // Set input callback
// AURenderCallbackStruct callbackStruct;
// callbackStruct.inputProc = recordingCallback;
// callbackStruct.inputProcRefCon = self;
// status = AudioUnitSetProperty(audioUnit,
// kAudioOutputUnitProperty_SetInputCallback,
// kAudioUnitScope_Global,
// kInputBus,
// &callbackStruct,
// sizeof(callbackStruct));
// checkStatus(status);
// Set output callback
AURenderCallbackStruct callbackStruct;
callbackStruct.inputProc = playbackCallback;
callbackStruct.inputProcRefCon = (__bridge void * _Nullable)(self);
AudioUnitSetProperty(_audioUnit,
kAudioUnitProperty_SetRenderCallback,
kAudioUnitScope_Global,
kOutputBus,
&callbackStruct,
sizeof(callbackStruct));
// Disable buffer allocation for the recorder (optional - do this if we want to pass in our own)
flag = 0;
status = AudioUnitSetProperty(_audioUnit,
kAudioUnitProperty_ShouldAllocateBuffer,
kAudioUnitScope_Output,
kInputBus,
&flag,
sizeof(flag));
//initialize the circular buffer
if(self.decodingBuffer == NULL)
self.decodingBuffer = malloc(sizeof(TPCircularBuffer));
if(!TPCircularBufferInit(self.decodingBuffer, 512*1024))
return NO;
// Initialise
status = AudioUnitInitialize(self.audioUnit);
AudioOutputUnitStart(self.audioUnit);
}
I found most of this by looking through github and from a tasty pixel

Answer 2

If the AVAudioSession is configured to use short buffers, you can use the RemoteIO Audio Unit to play received audio with low additional latency.
Check errors during audio configuration. Some iOS devices only support a 48 kHz sample rate, so you may need to resample your audio PCM data from 8 kHz to another rate.
RemoteIO only supports linear PCM, so you will need to first convert all your incoming 8-bit u-law PCM samples to 16-bit linear PCM format before storing them in a lock-free circular buffer.
You need to call AudioOutputUnitStart to start audio callbacks being called by the OS. Your code should not be calling these callbacks. They will be called by the OS.
AudioUnitRender is used for recording callbacks, not for playing audio. So you don't need to use it. Just fill the AudioBufferList buffers with the requested number of frames in the play callback.
Then you can use the play audio callback to check your circular buffer and pull the requested number of samples, if enough are available. You should not do any memory management (such as a free() call) inside this callback.