Splitting a fouriertransform in low, middle and high frequency intervals

Hi everybody.

I'm having some problems with a project I need to do for school. We have to make an analyzer for noise. We have to record some sound samples, and than do a fouriertransform of it, so that we can see the frequencies of the sound. That's working fine, but than the next step is to split the frequency spectrum into 3 intervals

• 1 with the low frequencies: 0-255 Hz
• 1 with the middle frequencies: 256Hz - 2kHz
• 1 with the higher frequencies 2KHz or higher.

So far this is my code. I used most of the code from this example.

    clear all
    clc
    Fs = 4000;                    % Sampling frequency
    T = 1/Fs;                     % Sample time
    L = 1000;                     % Length of signal
    t = (0:L-1)*T;                % Time vector

    % Record your voice for 5 seconds.
    recObj = audiorecorder(Fs, 24, 1);
    recordblocking(recObj,5);
    disp('End of Recording.');
    % Store data in double-precision array.
    myRecording = getaudiodata(recObj);

    FFT = 2^nextpow2(L); % Next power of 2 from length of y
    fourier = fft(myRecording);
    Y = fft(myRecording,FFT)/L;
    f = Fs/2*linspace(0,1,FFT/2+1);

    %Plot single-sided amplitude spectrum.
    plot(f,2*abs(Y(1:FFT/2+1)))
    title('Single-Sided Amplitude Spectrum of y(t)')
    axis([20 Inf -Inf Inf])
    xlabel('Frequency (Hz)')
    ylabel('|Y(f)|')

And this is an image from my plot when i let tones play of 500Hz, 800Hz and 1k5Hz

So that's working well

But I've no idea how I could split those things into intervals..

I don't want the whole solution, just a little push in the back, so I know what I've to do, so i can work on it again. Because now I'm just searching on the web for help or hints, with no result.

Help would be really appreciated :)

Answer 1

[Wayne King]

The key is to figure out the frequency spacing in the DFT, that uses simply what you are calling

 FFT = 2^nextpow2(L);

and the sampling frequency.

By the way, FFT is a bad variable name. You shouldn't name a variable FFT, how about NFFT?

Assuming you name the variable NFFT, the frequency spacing is

Fs/NFFT

Now you know your frequency increments and you should be able to roughly divide up your one-sided Y vector accordingly.

Answer 2

[Sammie]

Thanks, I was able to solve it now :)