Cannot render audio with VoiceProcessingIO AudioUnit on Mac - macos

Question

I'm currently attempting to use the VoiceProcessingIO AudioUnit to passthrough audio from the mic to another input device. I am using an userspace audio driver that proxies audio from output to input. I'm attempting to render the audio from the AudioUnit and save it into a ring buffer to then play it into that audio driver so it can be used as an input.
Here's a snippet of the code:
#define kNumber_Of_Channels 2
#define kLatency_Frame_Size 0
#define kRing_Buffer_Frame_Size ((65536 + kLatency_Frame_Size))
static Float32 *gMicrophoneRingBuffer;
static bool gMicrophoneRingBufferAllocated = false;
static AudioComponentInstance gAECAudioUnitInstance;
static AudioBufferList *gMicrophoneABL;
static bool gMicrophoneABLAllocated = false;
static OSStatus MicrophoneAECInputCallback(
void *inRefCon, AudioUnitRenderActionFlags *ioActionFlags,
const AudioTimeStamp *inTimeStamp, UInt32 inBusNumber,
UInt32 inNumberFrames, AudioBufferList *ioData) {
#pragma unused(inRefCon, ioData)
OSStatus theError;
// Render audio into audio buffer list
theError = AudioUnitRender(gAECAudioUnitInstance, ioActionFlags, inTimeStamp,
inBusNumber, inNumberFrames, gMicrophoneABL);
if (theError) {
NSLog(#"! Failed to render AEC audio unit\t\t%d", theError);
return theError;
}
// Copy audio buffer list into ring buffer
AudioBuffer mBuffer = gMicrophoneABL->mBuffers[0];
void *ioMainBuffer = mBuffer.mData;
UInt32 inIOBufferFrameSize = mBuffer.mDataByteSize / (sizeof(Float32) * 1);
UInt64 mSampleTime = inTimeStamp->mSampleTime;
UInt32 ringBufferFrameLocationStart = mSampleTime % kRing_Buffer_Frame_Size;
UInt32 firstPartFrameSize =
kRing_Buffer_Frame_Size - ringBufferFrameLocationStart;
UInt32 secondPartFrameSize = 0;
if (firstPartFrameSize >= inIOBufferFrameSize) {
firstPartFrameSize = inIOBufferFrameSize;
} else {
secondPartFrameSize = inIOBufferFrameSize - firstPartFrameSize;
}
memcpy(gMicrophoneRingBuffer + ringBufferFrameLocationStart * 1, ioMainBuffer,
firstPartFrameSize * 1 * sizeof(Float32));
memcpy(gMicrophoneRingBuffer,
(Float32 *)ioMainBuffer + firstPartFrameSize * 1,
secondPartFrameSize * 1 * sizeof(Float32));
return noErr;
}
static OSStatus MicrophoneOutputDeviceIOProc(AudioObjectID inDevice,
const AudioTimeStamp *inNow,
const AudioBufferList *inInputData,
const AudioTimeStamp *inInputTime,
AudioBufferList *outOutputData,
const AudioTimeStamp *inOutputTime,
void *__nullable inClientData) {
#pragma unused(inDevice, inNow, inInputData, inInputTime, inClientData)
AudioBuffer mBuffer = outOutputData->mBuffers[0];
UInt32 inIOBufferFrameSize = mBuffer.mDataByteSize / (sizeof(Float32) * 1);
UInt64 mSampleTime = inOutputTime->mSampleTime;
UInt32 ringBufferFrameLocationStart = mSampleTime % kRing_Buffer_Frame_Size;
UInt32 firstPartFrameSize =
kRing_Buffer_Frame_Size - ringBufferFrameLocationStart;
UInt32 secondPartFrameSize = 0;
if (firstPartFrameSize >= inIOBufferFrameSize) {
firstPartFrameSize = inIOBufferFrameSize;
} else {
secondPartFrameSize = inIOBufferFrameSize - firstPartFrameSize;
}
memcpy(mBuffer.mData,
gMicrophoneRingBuffer + ringBufferFrameLocationStart * 1,
firstPartFrameSize * 1 * sizeof(Float32));
memcpy((Float32 *)mBuffer.mData + firstPartFrameSize * 1,
gMicrophoneRingBuffer, secondPartFrameSize * 1 * sizeof(Float32));
return noErr;
}
static OSStatus getAudioObjectID(CFStringRef uid, AudioObjectID *result) {
AudioObjectPropertyAddress inAddress;
inAddress.mSelector = kAudioHardwarePropertyDeviceForUID;
UInt32 theSize = sizeof(AudioValueTranslation);
AudioValueTranslation theValue = {&uid, sizeof(CFStringRef), result,
sizeof(AudioObjectID)};
OSStatus theError = AudioObjectGetPropertyData(
kAudioObjectSystemObject, &inAddress, 0, NULL, &theSize, &theValue);
return theError;
}
static OSStatus
createAndStartAECAudioUnit(AudioObjectID microphoneId,
AudioDeviceIOProcID *outputIOProcId) {
gMicrophoneRingBuffer =
calloc(kRing_Buffer_Frame_Size * kNumber_Of_Channels, sizeof(Float32));
gMicrophoneRingBufferAllocated = true;
OSStatus theError;
// Find audio unit component
AudioComponentDescription auDesc;
auDesc.componentType = kAudioUnitType_Output;
auDesc.componentSubType = kAudioUnitSubType_VoiceProcessingIO;
auDesc.componentManufacturer = kAudioUnitManufacturer_Apple;
auDesc.componentFlags = 0;
auDesc.componentFlagsMask = 0;
AudioComponent component = AudioComponentFindNext(NULL, &auDesc);
// Create audio unit using component
theError = AudioComponentInstanceNew(component, &gAECAudioUnitInstance);
if (theError != noErr) {
NSLog(#"! Failed to create audio unit \t\t%d", theError);
return theError;
}
UInt32 one;
one = 1;
theError = AudioUnitSetProperty(gAECAudioUnitInstance,
kAudioOutputUnitProperty_EnableIO,
kAudioUnitScope_Input, 1, &one, sizeof(one));
if (theError != noErr) {
NSLog(#"! Failed to enable IO for input \t\t%d", theError);
return theError;
}
UInt32 zero;
zero = 0;
theError = AudioUnitSetProperty(
gAECAudioUnitInstance, kAudioOutputUnitProperty_EnableIO,
kAudioUnitScope_Output, 0, &zero, sizeof(zero));
if (theError != noErr) {
NSLog(#"! Failed to disable IO for output \t\t%d", theError);
return theError;
}
// Set current output device
theError = AudioUnitSetProperty(
gAECAudioUnitInstance, kAudioOutputUnitProperty_CurrentDevice,
kAudioUnitScope_Global, 0, &microphoneId, sizeof(AudioObjectID));
if (theError != noErr) {
NSLog(#"! Failed to set audio unit microphone device \t\t%d", theError);
return theError;
}
// Set input callback to pull the microphone data into ring buffer
AURenderCallbackStruct inputProc;
inputProc.inputProc = MicrophoneAECInputCallback;
theError = AudioUnitSetProperty(
gAECAudioUnitInstance, kAudioOutputUnitProperty_SetInputCallback,
kAudioUnitScope_Global, 0, &inputProc, sizeof(inputProc));
if (theError != noErr) {
NSLog(#"! Failed to set audio unit input callback \t\t%d", theError);
return theError;
}
// Get the size of the IO buffer(s)
UInt32 bufferSizeFrames;
UInt32 propertySize = sizeof(bufferSizeFrames);
theError = AudioUnitGetProperty(
gAECAudioUnitInstance, kAudioDevicePropertyBufferFrameSize,
kAudioUnitScope_Global, 0, &bufferSizeFrames, &propertySize);
if (theError != noErr) {
NSLog(#"! Failed to get buffer size frames \t\t%d", theError);
return theError;
}
// Initialize AudioBufferList
UInt32 bufferSizeBytes = bufferSizeFrames * sizeof(Float32);
UInt32 propsize =
offsetof(AudioBufferList, mBuffers[0]) + (sizeof(AudioBuffer) * 1);
gMicrophoneABL = (AudioBufferList *)malloc(propsize);
gMicrophoneABL->mNumberBuffers = streamDesc.mChannelsPerFrame;
for (UInt32 i = 0; i < gMicrophoneABL->mNumberBuffers; i++) {
gMicrophoneABL->mBuffers[i].mNumberChannels = streamDesc.mChannelsPerFrame;
gMicrophoneABL->mBuffers[i].mDataByteSize = bufferSizeBytes;
gMicrophoneABL->mBuffers[i].mData = malloc(bufferSizeBytes);
}
gMicrophoneABLAllocated = true;
// Initialize audio unit
theError = AudioUnitInitialize(gAECAudioUnitInstance);
if (theError != noErr) {
NSLog(#"! Failed to initialize audio unit \t\t%d", theError);
return theError;
}
// Start audio unit
theError = AudioOutputUnitStart(gAECAudioUnitInstance);
if (theError != noErr) {
NSLog(#"! Failed to start output of audio unit \t\t%d", theError);
return theError;
}
// Get microphone driver device id
AudioObjectID microphoneDriverDeviceId = kAudioObjectUnknown;
theError = getAudioObjectID(CFSTR(kMicrophone_Device_UID),
&microphoneDriverDeviceId);
if (theError != noErr) {
NSLog(#"! Failed to get microphone driver device UID\t\t%d", theError);
return theError;
}
// Create output proc
theError = AudioDeviceCreateIOProcID(microphoneDriverDeviceId,
MicrophoneOutputDeviceIOProc, NULL,
outputIOProcId);
if (theError != noErr) {
NSLog(#"! Failed to create output io proc\t\t%d", theError);
return theError;
}
// Starts the proxy process
theError = AudioDeviceStart(microphoneDriverDeviceId, *outputIOProcId);
if (theError != noErr) {
NSLog(#"! Failed to create output io proc\t\t%d", theError);
return theError;
}
return noErr;
}
So currently this code errors when it attempts to render the audio in the "MicrophoneAECInputCallback". This only happens when I change the AudioUnit component subtype to be "kAudioUnitSubType_VoiceProcessingIO" and not "kAudioUnitSubType_HALOutput", which leads me to believe that I'm just using the VoiceProcessingIO AudioUnit incorrectly.
I'm doing all this to perform echo cancellation on the microphone input to ultimately feed it into AVCaptureSession. Is there any alternative approaches any of you suggest or how should I go about fixing my usage of the VoiceProcessingIO AudioUnit on Mac.