AudioUnit Sound Generator gives me a chirp at the end of each generated tone

Question

I am creating an old school music emulator for the old GWBasic PLAY team. To do this, I have a tone generator and music player. Between each of the notes played, I get a tweet sound that throws things in. Both of my classes are listed below:

ToneGen.h

 #import <Foundation/Foundation.h> @interface ToneGen : NSObject @property (nonatomic) id delegate; @property (nonatomic) double frequency; @property (nonatomic) double sampleRate; @property (nonatomic) double theta; - (void)play:(float)ms; - (void)play; - (void)stop; @end 

ToneGen.m

 #import <AudioUnit/AudioUnit.h> #import "ToneGen.h" OSStatus RenderTone( void *inRefCon, AudioUnitRenderActionFlags *ioActionFlags, const AudioTimeStamp *inTimeStamp, UInt32 inBusNumber, UInt32 inNumberFrames, AudioBufferList *ioData); void ToneInterruptionListener(void *inClientData, UInt32 inInterruptionState); @interface ToneGen() @property (nonatomic) AudioComponentInstance toneUnit; @property (nonatomic) NSTimer *timer; - (void)createToneUnit; @end @implementation ToneGen @synthesize toneUnit = _toneUnit; @synthesize timer = _timer; @synthesize delegate = _delegate; @synthesize frequency = _frequency; @synthesize sampleRate = _sampleRate; @synthesize theta = _theta; - (id) init { self = [super init]; if (self) { self.sampleRate = 44100; self.frequency = 1440.0f; return self; } return nil; } - (void)play:(float)ms { [self play]; self.timer = [NSTimer scheduledTimerWithTimeInterval:(ms / 100) target:self selector:@selector(stop) userInfo:nil repeats:NO]; [[NSRunLoop mainRunLoop] addTimer:self.timer forMode:NSRunLoopCommonModes]; } - (void)play { if (!self.toneUnit) { [self createToneUnit]; // Stop changing parameters on the unit OSErr err = AudioUnitInitialize(self.toneUnit); if (err) DLog(@"Error initializing unit"); // Start playback err = AudioOutputUnitStart(self.toneUnit); if (err) DLog(@"Error starting unit"); } } - (void)stop { [self.timer invalidate]; self.timer = nil; if (self.toneUnit) { AudioOutputUnitStop(self.toneUnit); AudioUnitUninitialize(self.toneUnit); AudioComponentInstanceDispose(self.toneUnit); self.toneUnit = nil; } if(self.delegate && [self.delegate respondsToSelector:@selector(toneStop)]) { [self.delegate performSelector:@selector(toneStop)]; } } - (void)createToneUnit { AudioComponentDescription defaultOutputDescription; defaultOutputDescription.componentType = kAudioUnitType_Output; defaultOutputDescription.componentSubType = kAudioUnitSubType_DefaultOutput; defaultOutputDescription.componentManufacturer = kAudioUnitManufacturer_Apple; defaultOutputDescription.componentFlags = 0; defaultOutputDescription.componentFlagsMask = 0; // Get the default playback output unit AudioComponent defaultOutput = AudioComponentFindNext(NULL, &defaultOutputDescription); if (!defaultOutput) DLog(@"Can't find default output"); // Create a new unit based on this that we'll use for output OSErr err = AudioComponentInstanceNew(defaultOutput, &_toneUnit); if (err) DLog(@"Error creating unit"); // Set our tone rendering function on the unit AURenderCallbackStruct input; input.inputProc = RenderTone; input.inputProcRefCon = (__bridge void*)self; err = AudioUnitSetProperty(self.toneUnit, kAudioUnitProperty_SetRenderCallback, kAudioUnitScope_Input, 0, &input, sizeof(input)); if (err) DLog(@"Error setting callback"); // Set the format to 32 bit, single channel, floating point, linear PCM const int four_bytes_per_float = 4; const int eight_bits_per_byte = 8; AudioStreamBasicDescription streamFormat; streamFormat.mSampleRate = self.sampleRate; streamFormat.mFormatID = kAudioFormatLinearPCM; streamFormat.mFormatFlags = kAudioFormatFlagsNativeFloatPacked | kAudioFormatFlagIsNonInterleaved; streamFormat.mBytesPerPacket = four_bytes_per_float; streamFormat.mFramesPerPacket = 1; streamFormat.mBytesPerFrame = four_bytes_per_float; streamFormat.mChannelsPerFrame = 1; streamFormat.mBitsPerChannel = four_bytes_per_float * eight_bits_per_byte; err = AudioUnitSetProperty (self.toneUnit, kAudioUnitProperty_StreamFormat, kAudioUnitScope_Input, 0, &streamFormat, sizeof(AudioStreamBasicDescription)); if (err) DLog(@"Error setting stream format"); } @end OSStatus RenderTone( void *inRefCon, AudioUnitRenderActionFlags *ioActionFlags, const AudioTimeStamp *inTimeStamp, UInt32 inBusNumber, UInt32 inNumberFrames, AudioBufferList *ioData) { // Fixed amplitude is good enough for our purposes const double amplitude = 0.25; // Get the tone parameters out of the view controller ToneGen *toneGen = (__bridge ToneGen *)inRefCon; double theta = toneGen.theta; double theta_increment = 2.0 * M_PI * toneGen.frequency / toneGen.sampleRate; // This is a mono tone generator so we only need the first buffer const int channel = 0; Float32 *buffer = (Float32 *)ioData->mBuffers[channel].mData; // Generate the samples for (UInt32 frame = 0; frame < inNumberFrames; frame++) { buffer[frame] = sin(theta) * amplitude; theta += theta_increment; if (theta > 2.0 * M_PI) { theta -= 2.0 * M_PI; } } // Store the theta back in the view controller toneGen.theta = theta; return noErr; } void ToneInterruptionListener(void *inClientData, UInt32 inInterruptionState) { ToneGen *toneGen = (__bridge ToneGen *)inClientData; [toneGen stop]; } 

Music.h

 #import <Foundation/Foundation.h> @interface Music : NSObject - (void) play:(NSString *)music; - (void) stop; @end 

Music.m

 #import "Music.h" #import "ToneGen.h" @interface Music() @property (nonatomic, readonly) ToneGen *toneGen; @property (nonatomic, assign) int octive; @property (nonatomic, assign) int tempo; @property (nonatomic, assign) int length; @property (nonatomic, strong) NSData *music; @property (nonatomic, assign) int dataPos; @property (nonatomic, assign) BOOL isPlaying; - (void)playNote; @end @implementation Music @synthesize toneGen = _toneGen; - (ToneGen*)toneGen { if (_toneGen == nil) { _toneGen = [[ToneGen alloc] init]; _toneGen.delegate = self; } return _toneGen; } @synthesize octive = _octive; - (void)setOctive:(int)octive { // Sinity Check if (octive < 0) octive = 0; if (octive > 6) octive = 6; _octive = octive; } @synthesize tempo = _tempo; - (void)setTempo:(int)tempo { // Sinity Check if (tempo < 30) tempo = 30; if (tempo > 255) tempo = 255; _tempo = tempo; } @synthesize length = _length; - (void)setLength:(int)length { // Sinity Check if (length < 1) length = 1; if (length > 64) length = 64; _length = length; } @synthesize music = _music; @synthesize dataPos = _dataPos; @synthesize isPlaying = _isPlaying; - (id)init { self = [super init]; if (self) { self.octive = 4; self.tempo = 120; self.length = 1; return self; } return nil; } - (void) play:(NSString *)music { DLog(@"%@", music); self.music = [[music stringByReplacingOccurrencesOfString:@"+" withString:@"#"] dataUsingEncoding: NSASCIIStringEncoding]; self.dataPos = 0; self.isPlaying = YES; [self playNote]; } - (void)stop { self.isPlaying = NO; } - (void)playNote { if (!self.isPlaying) return; if (self.dataPos > self.music.length || self.music.length == 0) { self.isPlaying = NO; return; } unsigned char *data = (unsigned char*)[self.music bytes]; unsigned int code = (unsigned int)data[self.dataPos]; self.dataPos++; switch (code) { case 65: // A case 66: // B case 67: // C case 68: // D case 69: // E case 70: // F case 71: // G { // Peak at the next char to look for sharp or flat bool sharp = NO; bool flat = NO; if (self.dataPos < self.music.length) { unsigned int peak = (unsigned int)data[self.dataPos]; if (peak == 35) // # { self.dataPos++; sharp = YES; } else if (peak == 45) // - { self.dataPos++; flat = YES; } } // Peak ahead for a length changes bool look = YES; int count = 0; int newLength = 0; while (self.dataPos < self.music.length && look) { unsigned int peak = (unsigned int)data[self.dataPos]; if (peak >= 48 && peak <= 57) { peak -= 48; int n = (count * 10); if (n == 0) { n = 1; } newLength += peak * n; self.dataPos++; } else { look = NO; } } // Pick the note length int length = self.length; if (newLength != 0) { DLog(@"InlineLength: %d", newLength); length = newLength; } // Create the note string NSString *note = [NSString stringWithFormat:@"%c", code]; if (sharp) note = [note stringByAppendingFormat:@"#"]; else if (flat) note = [note stringByAppendingFormat:@"-"]; // Set the tone generator freq [self setFreq:[self getNoteNumber:note]]; // Play the note [self.toneGen play:(self.tempo / length)]; } break; case 76: // L (length) { bool look = YES; int newLength = 0; while (self.dataPos < self.music.length && look) { unsigned int peak = (unsigned int)data[self.dataPos]; if (peak >= 48 && peak <= 57) { peak -= 48; newLength = newLength * 10 + peak; self.dataPos++; } else { look = NO; } } self.length = newLength; DLog(@"Length: %d", self.length); [self playNote]; } break; case 79: // O (octive) { bool look = YES; int newOctive = 0; while (self.dataPos < self.music.length && look) { unsigned int peak = (unsigned int)data[self.dataPos]; if (peak >= 48 && peak <= 57) { peak -= 48; newOctive = newOctive * 10 + peak; self.dataPos++; } else { look = NO; } } self.octive = newOctive; DLog(@"Octive: %d", self.self.octive); [self playNote]; } break; case 84: // T (tempo) { bool look = YES; int newTempo = 0; while (self.dataPos < self.music.length && look) { unsigned int peak = (unsigned int)data[self.dataPos]; if (peak >= 48 && peak <= 57) { peak -= 48; newTempo = newTempo * 10 + peak; self.dataPos++; } else { look = NO; } } self.tempo = newTempo; DLog(@"Tempo: %d", self.self.tempo); [self playNote]; } break; default: [self playNote]; break; } } - (int)getNoteNumber:(NSString*)note { note = [note uppercaseString]; DLog(@"%@", note); if ([note isEqualToString:@"A"]) return 0; else if ([note isEqualToString:@"A#"] || [note isEqualToString:@"B-"]) return 1; else if ([note isEqualToString:@"B"] || [note isEqualToString:@"C-"]) return 2; else if ([note isEqualToString:@"C"] || [note isEqualToString:@"B#"]) return 3; else if ([note isEqualToString:@"C#"] || [note isEqualToString:@"D-"]) return 4; else if ([note isEqualToString:@"D"]) return 5; else if ([note isEqualToString:@"D#"] || [note isEqualToString:@"E-"]) return 6; else if ([note isEqualToString:@"E"] || [note isEqualToString:@"F-"]) return 7; else if ([note isEqualToString:@"F"] || [note isEqualToString:@"E#"]) return 8; else if ([note isEqualToString:@"F#"] || [note isEqualToString:@"G-"]) return 9; else if ([note isEqualToString:@"G"]) return 10; else if ([note isEqualToString:@"G#"]) return 11; } - (void)setFreq:(int)note { float a = powf(2, self.octive); float b = powf(1.059463, note); float freq = roundf((275.0 * a * b) / 10); self.toneGen.frequency = freq; } - (void)toneStop { [self playNote]; } @end 

To play a little customization , create a Music object and play ...

 [self.music play:@"T180 DF#A L2 A L4 O4 AA P4 F#F# P4 O3 D DF#A L2 A L4 O4 AA P4 GG P4 O3 C#C#EB L2 B L4 O4 BB P4 GG P4 O3 C#C#EB L2 B L4 O4 BB P4 F+F+ P4 O3 DDF#A L2 O4 D L4 O5 DD P4O4 AA P4 O3 DDF#A L2 O4 D L4 O5 DD P4O4 BB P4 EEG L8 B P8 ML B1 L4 MN G#A ML L3 O5 F#1L4 MN D O4 F# ML L2 F# MN L4 E ML L2 B MN L4 AD P8 D8 D4"]; 

Any idea on how to remove a chirp between notes?

Answer 1

I think the bit in which you stop the audio output between notes is the culprit:

 if (self.toneUnit) { AudioOutputUnitStop(self.toneUnit); AudioUnitUninitialize(self.toneUnit); AudioComponentInstanceDispose(self.toneUnit); self.toneUnit = nil; } 

Just leave the tone block active and you will have less twittering. You will need another way to generate silence, perhaps if the RenderTone continues to work, but will generate zero amplitude.

I was able to eliminate the minor tweets that remained if he, with a change in frequency, faded out the amplitude to zero, updated the frequency and disappeared again. This, of course, is something that the old PC speaker could not do (with the exception of a few people who quickly turned it on again), but with very fast fading, you can probably get the effect of the old school without tweets.

Here is my fading RenderTone function (currently using evil global variables):

 double currentFrequency=0; double currentSampleRate=0; double currentAmplitude=0; OSStatus RenderTone( void *inRefCon, AudioUnitRenderActionFlags *ioActionFlags, const AudioTimeStamp *inTimeStamp, UInt32 inBusNumber, UInt32 inNumberFrames, AudioBufferList *ioData) { // Fixed amplitude is good enough for our purposes const double amplitude = 0.5; // Get the tone parameters out of the view controller ToneGen *toneGen = (__bridge ToneGen *)inRefCon; double theta = toneGen.theta; BOOL fadingOut = NO; if ((currentFrequency != toneGen.frequency) || (currentSampleRate != toneGen.sampleRate)) { if (currentAmplitude > DBL_EPSILON) { fadingOut = YES; } else { currentFrequency = toneGen.frequency; currentSampleRate = toneGen.sampleRate; } } double theta_increment = 2.0 * M_PI * currentFrequency /currentSampleRate; // This is a mono tone generator so we only need the first buffer const int channel = 0; Float32 *buffer = (Float32 *)ioData->mBuffers[channel].mData; // Generate the samples for (UInt32 frame = 0; frame < inNumberFrames; frame++) { buffer[frame] = sin(theta) * currentAmplitude; //NSLog(@"amplitude = %f", currentAmplitude); theta += theta_increment; if (theta > 2.0 * M_PI) { theta -= 2.0 * M_PI; } if (fadingOut) { if (currentAmplitude > 0) { currentAmplitude -= 0.001; if (currentAmplitude < 0) currentAmplitude = 0; } } else { if (currentAmplitude < amplitude) { currentAmplitude += 0.001; if (currentAmplitude > amplitude) currentAmplitude = amplitude; } } } // Store the theta back in the view controller toneGen.theta = theta; return noErr; } 

Answer 2

This little chirp is usually an artifact of mathematics. The ear significantly analyzes the input signals in the frequency domain. For example, a stable sine wave at a frequency of 220 Hz will sound like A. However, when your sine wave is unstable, there are other frequencies that appear from abroad. In particular, you get some pop because of the very high frequency component of starting or stopping the sound all of a sudden.

The way I solved it in my synthesizer (in Javascript, not Obj-C, but the concept is the same here) is to fade out the sound in more than 300 samples or so, pay attention to and fade the sound from more than 300 samples or etc. There is no way to truly eliminate boundary effects other than no border at all, but even a small and imperceptible amount of attenuation will make the border effect invisible as well.