TPV Tracking Phase Vocoder

resynthesis = TPV.ar(chain, windowsize=1024, hopsize=512, maxpeaks=80, currentpeaks, freqmult=1.0, tolerance=4, noisefloor= 0.01, mul=1.0, add=0.0)

Implementation of the McAulay and Quatieri 1986 sinusoidal model, as described in:

McAulay, R. and Quatieri, T. (1986) Speech analysis/Synthesis based on a sinusoidal representation. IEEE Transactions on Acoustics, Speech, and Signal Processing 34(4): 744--754

chain [fft] - Audio input to track, which has been pre-analysed by the FFT UGen; see examples below for the expected FFT size

windowsize- Window size used for FFT

hopsize- hop size for FFT, typically half window size

maxpeaks- Absolute maximum number of allowed peaks to be detected in the spectrum

currentpeaks- Current number of allowed peaks to be detected in the spectrum

freqmult- Resynthesis parameter to change frequency; currently causes a gross multiplication of frequency of all sinusoidal components, no envelope compensation

tolerance- Search area for matching peaks; within tolerance spectral bins

noisefloor- Minimum magnitude for a candidate peak

b.free

//assumes hop of half fftsize, fine

b = Buffer.alloc(s,1024,1); //for sampling rates 44100 and 48000

//b = Buffer.alloc(s,2048,1); //for sampling rates 88200 and 96000

//d=Buffer.read(s,"sounds/a11wlk01.wav");

d=Buffer.read(s,"sounds/break");

(

{

var in, fft, output;

in= SoundIn.ar(0); //PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);

fft = FFT(b, in, wintype:1);

output=TPV.ar(fft, 1024, 512, 70,MouseX.kr(1,70), MouseY.kr(0.25,3),4,0.2); 

//Out.ar(0,Pan2.ar(output)); 

output

}.play

)

(

{

var in, fft, output;

in= PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);

fft = FFT(b, in, wintype:1);

output=TPV.ar(fft, 1024, 512, 50,50,1.0,MouseX.kr(0.1,100),MouseY.kr(-20,40).dbamp); 

//Out.ar(0,Pan2.ar(output)); 

output

}.play

)

(

{

var in, fft, output;

in= PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);

fft = FFT(b, in);

output=TPV.ar(fft, 1024, 512, 1,1); 

output

//Out.ar(0,output); 

}.plot(0.05, s, nil, -1.5, 1.5);

)

//residual

b = Buffer.alloc(s,1024,1); //for sampling rates 44100 and 48000

d=Buffer.read(s,"sounds/break");

(

{

var in, fft, output;

var input,inputAmp,threshhold,gate;

in= SoundIn.ar(0); //PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);

//inputAmp = Amplitude.kr(input);

//threshhold = 0.001; // noise gating threshold

//gate = Lag.kr(inputAmp > threshhold, 0.01);

//in= (input * gate);

fft = FFT(b, in, wintype:1);

output=TPV.ar(fft, 1024, 512, 50,50,1.0,4,0.2); 

//Out.ar(0,Pan2.ar(output)); 

//delay by 512 samples for output, then phase measurement is from centre of frame, so  

output - DelayN.ar(in,0.1, (512+512)/44100)

}.play

)