MATLAB: Dividing speech signal into short-time segments

Hi all. I created a speech signal in Matlab. I need to divide speech signal into short-time segments with lengths 150 samples. Later I will process each segment to determine if it is voiced or unvoiced, and pitch period for voiced speechs. Here is the info:

Best Answer

If your sound signal is a vector, try to use reshape. For example from the info I understand that your sound signal would be a vector with 72000 elements which you want to divide in groups of 150.

SeperatedSoundSignal = reshape(soundSignal,150,72000/150);

This will convert your vector into 72000/150 columns which have each 150 elements/samples.

Kind regards, Baltam

Related Solutions

MATLAB: Segmenting a speech signal and adding noise

If you have the Signal Processing Toolbox, you can use buffer(). Otherwise you can use reshape(), but buffer() allows you to easily overlap the frames.

For example, I'll assume that the sampling frequency is 40 kHz, so that in this case 800 samples is 0.020

   t = 0:1/40e3:10-(1/40e3);
   x = cos(2*pi*2000*t)+sin(2*pi*4000*t);
   x = x(:);
   Y = buffer(x,800);

The above gives you 500 frames of 800 samples each with no overlap.

Now if you want a 0.005 overlap for each frame, that is 200 samples.

    Y = buffer(x,800,200);

Then you can create a matrix of noise the same size as Y and add that noise matrix to your signal matrix.

MATLAB: How do you find average power

Calculating the average power of your discrete signal according to the equation in your attached figure is done as:

P = sum(x.^2)/L;

P = mean(x.^2);

P = rms(x)^2;

Best Answer

Related Solutions

MATLAB: Segmenting a speech signal and adding noise

MATLAB: How do you find average power

Related Question