SHOW:
|
|
- or go back to the newest paste.
| 1 | function y = sinemodel(x, w, N, t, fs) | |
| 2 | % Authors: J. Bonada, X. Serra, X. Amatriain, A. Loscos | |
| 3 | % Analysis/synthesis of a sound using the sinusoidal model | |
| 4 | % x: input sound, w: analysis window (odd size), N: FFT size, | |
| 5 | % t: threshold in negative dB, y: output sound | |
| 6 | % | |
| 7 | %-------------------------------------------------------------------------- | |
| 8 | % This source code is provided without any warranties as published in | |
| 9 | % DAFX book 2nd edition, copyright Wiley & Sons 2011, available at | |
| 10 | % http://www.dafx.de. It may be used for educational purposes and not | |
| 11 | % for commercial applications without further permission. | |
| 12 | %-------------------------------------------------------------------------- | |
| 13 | ||
| 14 | M = length(w); % analysis window size | |
| 15 | Ns= 1024; % FFT size for synthesis (even) | |
| 16 | H = 256; % analysis/synthesishop size | |
| 17 | N2= N/2+1; % size of positive spectrum | |
| 18 | soundlength = length(x); % length of input sound array | |
| 19 | hNs = Ns/2; % half synthesis window size | |
| 20 | hM = (M-1)/2; % half analysis window size | |
| 21 | pin = max(H+1,1+hM); % initialize sound pointer to middle of analysis window | |
| 22 | pend = soundlength-max(H,hM); % last sample to start a frame | |
| 23 | fftbuffer = zeros(N,1); % initialize buffer for FFT | |
| 24 | y = zeros(soundlength,1); % initialize output array | |
| 25 | w = w/sum(w); % normalize analysis window | |
| 26 | sw = zeros(Ns,1); | |
| 27 | ow = triang(2*H-1); % overlapping window | |
| 28 | ovidx = Ns/2+1-H+1:Ns/2+H; % overlap indexes | |
| 29 | sw(ovidx) = ow(1:2*H-1); | |
| 30 | bh = blackmanharris(Ns); % synthesis window | |
| 31 | bh = bh ./ sum(bh); % normalize synthesis window | |
| 32 | sw(ovidx) = sw(ovidx) ./ bh(ovidx); | |
| 33 | k = [0:1:N]; | |
| 34 | freq = fs*k/N; | |
| 35 | minf0 = 100; | |
| 36 | maxf0 = 1200; | |
| 37 | ef0max = 100; | |
| 38 | % while pin<pend | |
| 39 | %-----analysis-----% | |
| 40 | xw = x(pin-hM:pin+hM).*w(1:M); % window the input sound | |
| 41 | fftbuffer(:) = 0; % reset buffer | |
| 42 | fftbuffer(1:(M+1)/2) = xw((M+1)/2:M); % zero-phase window in fftbuffer | |
| 43 | fftbuffer(N-(M-1)/2+1:N) = xw(1:(M-1)/2); | |
| 44 | X = fft(fftbuffer); % compute the FFT | |
| 45 | mX = 20*log10(abs(X(1:N2))); % magnitude spectrum of positive frequencies | |
| 46 | pX = unwrap(angle(X(1:N/2+1))); % unwrapped phase spectrum | |
| 47 | ploc = 1 + find((mX(2:N2-1)>t) .* (mX(2:N2-1)>mX(3:N2)) ... | |
| 48 | .* (mX(2:N2-1)>mX(1:N2-2))); % find peaks | |
| 49 | [ploc,pmag,pphase] = peakinterp(mX,pX,ploc); % refine peak values | |
| 50 | f0 = f0detection(mX, fs, ploc, pmag, ef0max, minf0, maxf0); | |
| 51 | f0yin = f0detectionyin(x, fs, 1000, minf0, maxf0) ; | |
| 52 | f0 | |
| 53 | f0yin | |
| 54 | - | subplot(2,1,1); |
| 54 | + | % subplot(2,1,1); |
| 55 | plot(freq(1:N2),mX); | |
| 56 | hold on; | |
| 57 | plot(fs*ploc/N,pmag,'rX'); | |
| 58 | plot(f0,mX(round(f0*N/fs)),'b*'); | |
| 59 | plot(f0yin,mX(round(f0yin*N/fs)),'g*'); | |
| 60 | hold off; | |
| 61 | % subplot(2,1,2); | |
| 62 | % plot(f0,mX(round(f0*N/fs)),'X'); | |
| 63 | % %-----synthesis-----% | |
| 64 | % plocs = (ploc-1)*Ns/N; % adapt peak locations to synthesis FFT | |
| 65 | % Y = genspecsines(plocs,pmag,pphase,Ns); % generate spec sines | |
| 66 | % yw = fftshift(real(ifft(Y))); % time domain of sinusoids | |
| 67 | % %(pin-hNs:pin+hNs-1) | |
| 68 | % y(pin-hNs:pin+hNs-1) = y(pin-hNs:pin+hNs-1) + sw.*yw(1:Ns); % overlap-add | |
| 69 | % pin = pin+H; % advance the sound pointer | |
| 70 | - | end |
| 70 | + | end |
| 71 | ||
| 72 |
