FeatureSpectralMfccs.m

%computes the MFCCs from the magnitude spectrum (see Slaney)
%> called by ::ComputeFeature
%>
%> @param X: spectrogram (dimension FFTLength X Observations)
%> @param f_s: sample rate of audio data (unused)
%>
%> @retval vmfcc mel frequency cepstral coefficients
% ======================================================================
function [vmfcc] = FeatureSpectralMfccs(X, f_s)

    iNumCoeffs = 13;
    
    % allocate memory
    vmfcc = zeros(iNumCoeffs, size(X, 2));

    % create filters
    H = ToolMfccFb(size(X, 1), f_s);
    
    % create transformation matrix
    T = GenerateDctMatrix_I (size(H, 1), iNumCoeffs);
 
    for n = 1:size(X,2)
        % compute the mel spectrum
        X_Mel = log10(H * X(:, n) + 1e-20);

        % calculate the mfccs
        vmfcc(:, n) = T * X_Mel;
    end
end


%> see function mfcc.m from Slaneys Auditory Toolbox
function [T] = GenerateDctMatrix_I (iNumBands, iNumCepstralCoeffs)
    T = cos((0:(iNumCepstralCoeffs-1))' * ...
            (2*(0:(iNumBands-1))+1) * pi/2/iNumBands);
            
    T  = T/sqrt(iNumBands/2);
    T(1, :)  = T(1, :) * sqrt(2)/2;
end