www.pudn.com > audioProcessingtoolbox.rar > fftspectrum.m

function [freq, Pyy] = fftspectrum(frame, fs, fftmethod, fspace) 
% FFTSPECTRUM Short term FFT to get magnitude spectrum information 
%    Usage : [freq, Pyy] = fftspectrum(frame, fs, fftmethod, fspace) 
%    frame means frame vector from some wavefile. 
%    fs means sample rate of the wavefile. 
%    fftmethod presents two different extraction rules. Its value is either 1 or 2. 
%    fspace means frequency spacing. 
%    freq represents frequency components 
%    Pyy represents log magnitudes corresponding to frequency components. 
% 
%    Cheng-Yuan Lin 2003, January, 22. 
 
if nargin<4,  
   fspace = 0: 20 : (fs/2);  %20 Hz spacing. 
end; 
if nargin<3, fftmethod = 1; end; 
 
switch fftmethod, 
case 1, 
   % general version 
   frameSize = length(frame); 
   Y    = fft(frame); 
   half = ceil(frameSize/2); 
   Pyy  = 20*log10(abs(Y)); 
   Pyy  = Pyy(1:half)';  %fft spectrum is symmetric. 
   %freq = fs/frameSize*(1:frameSize); 
   %freq = freq(1:half); 
   spectfactor = (fs/2)/length(Pyy); 
   freq = [1: spectfactor : (fs/2)]; 
case 2, 
   % Using interpolation version 
   frameSize = length(frame); 
   lf = length(fspace); 
   f = fft(frame, max(128, power(2, round(log2(frameSize))))); 
   f = 20*log10(1e-10 + abs(f(2:round(length(f)/2)))); 
   Pyy = interp1(linspace(0, fs/2, length(f)), f, fspace); 
   freq = (fs/2)/length(Pyy)*(1:length(Pyy))'; 
end;