kode54
/
cog
2
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[FreeSurround] The rest of the implementation 

This is the code that actually needs to be added to make it process
audio. This insertion makes the whole app crash when processing audio at
all. Weirdly, simply reverting these two files makes the audio code work
again. I can't explain it.

Also, commenting out CMAudioFormatDescriptionCreate makes it work, too.
There's something weird going on with that function.

Signed-off-by: Christopher Snowhill <kode54@gmail.com>
main
Christopher Snowhill 2022-07-10 15:21:09 -07:00
parent 85db95befd
commit 11413ef3db
2 changed files with 73 additions and 45 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
Audio/Output/OutputAVFoundation.h
View File

@ -35,6 +35,8 @@ using std::atomic_long;
@class OutputNode; @class OutputNode;
@class FSurroundFilter;
@interface OutputAVFoundation : NSObject { @interface OutputAVFoundation : NSObject {
OutputNode *outputController; OutputNode *outputController;
@ -58,6 +60,7 @@ using std::atomic_long;
BOOL eqInitialized; BOOL eqInitialized;
BOOL streamFormatStarted; BOOL streamFormatStarted;
BOOL streamFormatChanged;
double secondsHdcdSustained; double secondsHdcdSustained;
@ -106,10 +109,15 @@ using std::atomic_long;
BOOL enableHrtf; BOOL enableHrtf;
HeadphoneFilter *hrtf; HeadphoneFilter *hrtf;
BOOL enableFSurround;
FSurroundFilter *fsurround;
float inputBuffer[2048 * 32]; // 2048 samples times maximum supported channel count float *samplePtr;
float hrtfBuffer[2048 * 2]; float inputBuffer[4096 * 32]; // 4096 samples times maximum supported channel count
float eqBuffer[2048 * 32]; float fsurroundBuffer[4096 * 6];
float hrtfBuffer[4096 * 2];
float eqBuffer[4096 * 32];
#ifdef OUTPUT_LOG #ifdef OUTPUT_LOG
FILE *_logFile; FILE *_logFile;

104
Audio/Output/OutputAVFoundation.m
View File

@ -19,6 +19,8 @@
#import "r8bstate.h" #import "r8bstate.h"
#import "FSurroundFilter.h"
extern void scale_by_volume(float *buffer, size_t count, float volume); extern void scale_by_volume(float *buffer, size_t count, float volume);
static NSString *CogPlaybackDidBeginNotficiation = @"CogPlaybackDidBeginNotficiation"; static NSString *CogPlaybackDidBeginNotficiation = @"CogPlaybackDidBeginNotficiation";
@ -46,22 +48,20 @@ static void clearBuffers(AudioBufferList *ioData, size_t count, size_t offset) {
} }
static OSStatus eqRenderCallback(void *inRefCon, AudioUnitRenderActionFlags *ioActionFlags, const AudioTimeStamp *inTimeStamp, UInt32 inBusNumber, UInt32 inNumberFrames, AudioBufferList *ioData) { static OSStatus eqRenderCallback(void *inRefCon, AudioUnitRenderActionFlags *ioActionFlags, const AudioTimeStamp *inTimeStamp, UInt32 inBusNumber, UInt32 inNumberFrames, AudioBufferList *ioData) {
if(inNumberFrames > 1024 || !inRefCon) { if(inNumberFrames > 4096 || !inRefCon) {
clearBuffers(ioData, inNumberFrames, 0); clearBuffers(ioData, inNumberFrames, 0);
return 0; return 0;
} }
OutputAVFoundation *_self = (__bridge OutputAVFoundation *)inRefCon; OutputAVFoundation *_self = (__bridge OutputAVFoundation *)inRefCon;
fillBuffers(ioData, &_self->inputBuffer[0], inNumberFrames, 0); fillBuffers(ioData, _self->samplePtr, inNumberFrames, 0);
return 0; return 0;
} }
- (int)renderInput { - (int)renderInput:(int)amountToRead toBuffer:(float *)buffer {
int amountToRead, amountRead = 0; int amountRead = 0;
amountToRead = 1024;
float visAudio[amountToRead]; // Chunk size float visAudio[amountToRead]; // Chunk size
@ -71,7 +71,7 @@ static OSStatus eqRenderCallback(void *inRefCon, AudioUnitRenderActionFlags *ioA
return 0; return 0;
} }
AudioChunk *chunk = [outputController readChunk:1024]; AudioChunk *chunk = [outputController readChunk:amountToRead];
int frameCount = (int)[chunk frameCount]; int frameCount = (int)[chunk frameCount];
AudioStreamBasicDescription format = [chunk format]; AudioStreamBasicDescription format = [chunk format];
@ -132,12 +132,12 @@ static OSStatus eqRenderCallback(void *inRefCon, AudioUnitRenderActionFlags *ioA
location:@"pre downmix"]; location:@"pre downmix"];
#endif #endif
// It should be fine to request up to double, we'll only get downsampled // It should be fine to request up to double, we'll only get downsampled
float outputBuffer[2048 * newFormat.mChannelsPerFrame]; float outputBuffer[amountToRead * newFormat.mChannelsPerFrame];
const float *outputPtr = (const float *)[samples bytes]; const float *outputPtr = (const float *)[samples bytes];
if(r8bstate) { if(r8bstate) {
size_t inDone = 0; size_t inDone = 0;
[currentPtsLock lock]; [currentPtsLock lock];
size_t framesDone = r8bstate_resample(r8bstate, outputPtr, frameCount, &inDone, &outputBuffer[0], 2048); size_t framesDone = r8bstate_resample(r8bstate, outputPtr, frameCount, &inDone, &outputBuffer[0], amountToRead);
[currentPtsLock unlock]; [currentPtsLock unlock];
if(!framesDone) return 0; if(!framesDone) return 0;
frameCount = (int)framesDone; frameCount = (int)framesDone;
@ -213,7 +213,7 @@ static OSStatus eqRenderCallback(void *inRefCon, AudioUnitRenderActionFlags *ioA
[visController postVisPCM:&visAudio[0] amount:frameCount]; [visController postVisPCM:&visAudio[0] amount:frameCount];
} }
cblas_scopy((int)(frameCount * newFormat.mChannelsPerFrame), outputPtr, 1, &inputBuffer[0], 1); cblas_scopy((int)(frameCount * newFormat.mChannelsPerFrame), outputPtr, 1, &buffer[0], 1);
amountRead = frameCount; amountRead = frameCount;
} else { } else {
return 0; return 0;
@ -237,7 +237,7 @@ static OSStatus eqRenderCallback(void *inRefCon, AudioUnitRenderActionFlags *ioA
volumeScale *= eqPreamp; volumeScale *= eqPreamp;
} }
scale_by_volume(&inputBuffer[0], amountRead * newFormat.mChannelsPerFrame, volumeScale); scale_by_volume(&buffer[0], amountRead * newFormat.mChannelsPerFrame, volumeScale);
return amountRead; return amountRead;
} }
@ -600,25 +600,37 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
- (void)updateStreamFormat { - (void)updateStreamFormat {
/* Set the channel layout for the audio queue */ /* Set the channel layout for the audio queue */
streamFormatChanged = YES;
uint32_t channels = realStreamFormat.mChannelsPerFrame;
uint32_t channelConfig = realStreamChannelConfig;
if(enableFSurround && channels == 2 && channelConfig == AudioConfigStereo) {
fsurround = [[FSurroundFilter alloc] initWithSampleRate:realStreamFormat.mSampleRate];
channels = [fsurround channelCount];
channelConfig = [fsurround channelConfig];
} else {
fsurround = nil;
}
/* Apple's audio processor really only supports common 1-8 channel formats */ /* Apple's audio processor really only supports common 1-8 channel formats */
if(enableHrtf || realStreamFormat.mChannelsPerFrame > 8 || ((realStreamChannelConfig & ~(AudioConfig6Point1|AudioConfig7Point1)) != 0)) { if(enableHrtf || channels > 8 || ((channelConfig & ~(AudioConfig6Point1|AudioConfig7Point1)) != 0)) {
NSURL *presetUrl = [[NSBundle mainBundle] URLForResource:@"SADIE_D02-96000" withExtension:@"mhr"]; NSURL *presetUrl = [[NSBundle mainBundle] URLForResource:@"SADIE_D02-96000" withExtension:@"mhr"];
hrtf = [[HeadphoneFilter alloc] initWithImpulseFile:presetUrl forSampleRate:realStreamFormat.mSampleRate withInputChannels:realStreamFormat.mChannelsPerFrame withConfig:realStreamChannelConfig]; hrtf = [[HeadphoneFilter alloc] initWithImpulseFile:presetUrl forSampleRate:realStreamFormat.mSampleRate withInputChannels:channels withConfig:channelConfig];
streamFormat = realStreamFormat; channels = 2;
streamFormat.mChannelsPerFrame = 2; channelConfig = AudioChannelSideLeft | AudioChannelSideRight;
streamFormat.mBytesPerFrame = sizeof(float) * 2;
streamFormat.mFramesPerPacket = 1;
streamFormat.mBytesPerPacket = streamFormat.mBytesPerFrame;
streamChannelConfig = AudioChannelSideLeft | AudioChannelSideRight;
} else { } else {
hrtf = nil; hrtf = nil;
streamFormat = realStreamFormat;
streamChannelConfig = realStreamChannelConfig;
} }
streamFormat = realStreamFormat;
streamFormat.mChannelsPerFrame = channels;
streamFormat.mBytesPerFrame = sizeof(float) * channels;
streamFormat.mFramesPerPacket = 1;
streamFormat.mBytesPerPacket = sizeof(float) * channels;
streamChannelConfig = channelConfig;
AudioChannelLayoutTag tag = 0; AudioChannelLayoutTag tag = 0;
@ -683,7 +695,7 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
asbd.mBytesPerPacket = sizeof(float); asbd.mBytesPerPacket = sizeof(float);
asbd.mFramesPerPacket = 1; asbd.mFramesPerPacket = 1;
UInt32 maximumFrames = 1024; UInt32 maximumFrames = 4096;
AudioUnitSetProperty(_eq, kAudioUnitProperty_MaximumFramesPerSlice, kAudioUnitScope_Global, 0, &maximumFrames, sizeof(maximumFrames)); AudioUnitSetProperty(_eq, kAudioUnitProperty_MaximumFramesPerSlice, kAudioUnitScope_Global, 0, &maximumFrames, sizeof(maximumFrames));
AudioUnitSetProperty(_eq, kAudioUnitProperty_StreamFormat, AudioUnitSetProperty(_eq, kAudioUnitProperty_StreamFormat,
@ -714,11 +726,10 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
CMSampleBufferRef sampleBuffer = nil; CMSampleBufferRef sampleBuffer = nil;
OSStatus err = CMAudioSampleBufferCreateReadyWithPacketDescriptions(kCFAllocatorDefault, blockBuffer, audioFormatDescription, samplesRendered, outputPts, nil, &sampleBuffer); OSStatus err = CMAudioSampleBufferCreateReadyWithPacketDescriptions(kCFAllocatorDefault, blockBuffer, audioFormatDescription, samplesRendered, outputPts, nil, &sampleBuffer);
CFRelease(blockBuffer);
if(err != noErr) { if(err != noErr) {
CFRelease(blockBuffer);
return nil; return nil;
} }
CFRelease(blockBuffer);
return sampleBuffer; return sampleBuffer;
} }
@ -730,18 +741,27 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
status = CMBlockBufferCreateEmpty(kCFAllocatorDefault, 0, 0, &blockListBuffer); status = CMBlockBufferCreateEmpty(kCFAllocatorDefault, 0, 0, &blockListBuffer);
if(status != noErr || !blockListBuffer) return 0; if(status != noErr || !blockListBuffer) return 0;
int inputRendered; int inputRendered = 0;
int bytesRendered = 0;
do { do {
inputRendered = [self renderInput]; int maxToRender = MIN(4096 - inputRendered, 512);
int rendered = [self renderInput:maxToRender toBuffer:(float *)(((uint8_t *)inputBuffer) + bytesRendered)];
inputRendered += rendered;
bytesRendered += rendered * newFormat.mBytesPerPacket;
if(streamFormatChanged) {
streamFormatChanged = NO;
if(rendered < maxToRender) {
break;
}
}
if([self processEndOfStream]) break; if([self processEndOfStream]) break;
} while(!inputRendered); } while(inputRendered < 4096);
float tempBuffer[2048 * 32]; float tempBuffer[4096 * 32];
int samplesRenderedTotal = 0; int samplesRenderedTotal = 0;
for(size_t i = 0; i < 2;) { for(size_t i = 0; i < 2;) {
float *samplePtr;
int samplesRendered; int samplesRendered;
if(i == 0) { if(i == 0) {
@ -750,12 +770,14 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
continue; continue;
} }
[currentPtsLock lock]; [currentPtsLock lock];
samplesRendered = r8bstate_flush(r8bold, &tempBuffer[0], 2048); samplesRendered = r8bstate_flush(r8bold, &tempBuffer[0], 4096);
[currentPtsLock unlock]; [currentPtsLock unlock];
if(!samplesRendered) { if(!samplesRendered) {
r8bstate_delete(r8bold); r8bstate_delete(r8bold);
r8bold = NULL; r8bold = NULL;
r8bDone = YES; r8bDone = YES;
++i;
continue;
} }
samplePtr = &tempBuffer[0]; samplePtr = &tempBuffer[0];
} else { } else {
@ -770,6 +792,11 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
} }
if(samplesRendered) { if(samplesRendered) {
if(fsurround) {
[fsurround process:samplePtr output:&fsurroundBuffer[0] count:samplesRendered];
samplePtr = &fsurroundBuffer[0];
}
if(hrtf) { if(hrtf) {
[hrtf process:samplePtr sampleCount:samplesRendered toBuffer:&hrtfBuffer[0]]; [hrtf process:samplePtr sampleCount:samplesRendered toBuffer:&hrtfBuffer[0]];
samplePtr = &hrtfBuffer[0]; samplePtr = &hrtfBuffer[0];
@ -784,7 +811,7 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
ioData->mNumberBuffers = channels; ioData->mNumberBuffers = channels;
for(size_t i = 0; i < channels; ++i) { for(size_t i = 0; i < channels; ++i) {
ioData->mBuffers[i].mData = &eqBuffer[1024 * i]; ioData->mBuffers[i].mData = &eqBuffer[4096 * i];
ioData->mBuffers[i].mDataByteSize = samplesRendered * sizeof(float); ioData->mBuffers[i].mDataByteSize = samplesRendered * sizeof(float);
ioData->mBuffers[i].mNumberChannels = 1; ioData->mBuffers[i].mNumberChannels = 1;
} }
@ -799,7 +826,7 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
timeStamp.mSampleTime += ((double)samplesRendered) / streamFormat.mSampleRate; timeStamp.mSampleTime += ((double)samplesRendered) / streamFormat.mSampleRate;
for(int i = 0; i < channels; ++i) { for(int i = 0; i < channels; ++i) {
cblas_scopy(samplesRendered, &eqBuffer[1024 * i], 1, samplePtr + i, channels); cblas_scopy(samplesRendered, &eqBuffer[4096 * i], 1, samplePtr + i, channels);
} }
} }
} }
@ -840,14 +867,7 @@ current_device_listener(AudioObjectID inObjectID, UInt32 inNumberAddresses, cons
} }
} else { } else {
samplesRenderedTotal += samplesRendered; samplesRenderedTotal += samplesRendered;
if(!samplesRendered || samplesRenderedTotal >= 1024) { ++i;
++i;
} else {
do {
inputRendered = [self renderInput];
if([self processEndOfStream]) break;
} while(!inputRendered);
}
} }
} }