Comparing two audio signals (guitar chords recognizer)

I want to compare two audio signals and check their similarities however it seems i cannot compare them.. this is supposedly my process.

• > input audio signal (1 strum on guitar)
• > read the data and convert to frequency domain
• > get the raw audio file > read > convert to frequency domain
• > check their similarities

function guitartrial
    recObj = audiorecorder;
    disp('Start speaking.')
    recordblocking(recObj, 4);
    disp('End of Recording.');
    myRecording = getaudiodata(recObj);
    wavwrite(myRecording,'chordread.wav');
    [x, Fs] = wavread('chordread.wav');
    [x1, Fs1] = wavread('chordA.wav');
    Y = fft(abs(x));
    Z = fft(abs(x1));
    subplot(3,1,1), plot (Y);
    subplot(3,1,2), plot (Z);
    [C1, lag1] = xcorr(Y,Z);
    subplot(3,1,3), plot(lag1/Fs,C1);
    ylabel('Amplitude'); grid on
    title('Cross-correlation between Template 1 and Signal')
end

whenever i run the program i get this..

It seems wrong to me.. Im new in matlab programming so If someone can help me I would be very grateful. Is my process wrong? how can I get the percentage of similarity? is there another method that is best suited for my application? thank you very much

Answer 1

[Suneesh]

If your confusion is regarding the first and second plots then that is because yo uare plotting the complex values returned by FFT. Try:

    subplot(3,1,1), plot(abs(Y));
    subplot(3,1,2), plot(abs(Z));

Also, in

Y = fft(abs(x));

The abs(x) is not required since audio signals are real signals.

[Phantom.i7]

[C1, lag1] = xcorr(Y,Z);

is that correct or should i make it

[C1, lag1] = xcorr(abs(Y),abs(Z));

[Suneesh]

If you need to look for similarities in the magnitude of the two FFT results then

[C1, lag1] = xcorr(abs(Y),abs(Z));

is correct. This does not consider the phases which is lost by the ABS operation. Also, if the samples are of different lengths then make sure both are normalized.

Answer 2

[KIRAN SL]

what is the correlation plot means?