welch - Welch spectrum estimation
The time series is splitted with overlapping, each segment is windowed by hamming, then FFT transforms are done.
H represents not a power spectrum but an amplitude spectrum.
x=rand(1,10000,'normal'); hz=iir(8,'bp','butt',[0.1,0.25],[1E-5,1E-5]); y=flts(x,hz); [H1,W1]=welch(y,1024); plot2d(W1 ./(2 .*%pi),H1,style=3); xclick(); [H2,W2]=welch(y,256); plot2d(W2 ./(2 .*%pi),H2,style=2); // End of wlech demo