Frame by frame freqeuncy analysis of voice.

below code for frame by frame analysis of wav file , Seniors Go through it and answer my question please , question is written in the last.

[data,fs]= audioread('filename.wav);

fd=0.025 % duration selected from 5 sec

f_size= round(fd*fs);% how many samples in a frame

n_f=floor(length(data)/f_size) %how many frames are there

temp=0; %temporary value

for i= 1: n_f

frames(i,:) =data (temp+1: temp+f_size);

temp=temp+f_size;

end

plot (frame(1,:))%plot first frame or any you want give there numbe .

now.my question is that how can I read frequency of that particular frame or that duration of 0.025 ms? please help me for it .

Answer 1

[Bjorn Gustavsson]

Have a look at the documentation for the spectrogram function, that should do what you want. 0.025 is a rather short time so perhaps you can get something from calculating the instantaneous frequency - so have a look at that (and first the Hilbert transform too), this is likely a rough suggestion since we expect frequencies way higher than 20 Hz.

HTH

[Yasir Ali]

So any other way to do it ?

can I get frequency for 1/4 part of a Second ?Dear Kindly help me to get the code for Such program in which we can get frequency for micro seconds , or in different part of a second . as I have five Seconds wav file I want to get frequency of different part of 5 seconds .

[Bjorn Gustavsson]

Have you looked at the help for spectrogram? What could you use that function for to solve your problem? Did you look at the documentation for hilbert-transform and instantaneous frequency? Did that provide steps towards a solution of your problem?

Take a look at spectrogram.

[Yasir Ali]

no I didnt go before for hilbert transform and instantaneous frequency I will go through it .

[Yasir Ali]

is there anyother way in which I can get solution of my problem easily I just want to get frequency in different parts of second e.g from 2 seconds audio file I read Some frames or 1/4 part of a second and want to plot there freqeuncy spectrum as I above mentioned the way I extract frames of voice as same way or little bit change I want the Solution ay experts n,Senior help me to get the code for my Problem.

[Bjorn Gustavsson]

Yes. The SPECTROGRAM function does what you need - perhaps not what you think you want, but what you need. Let us look at the help text for spectrogram:

 SPECTROGRAM Spectrogram using a Short-Time Fourier Transform (STFT).
    S = SPECTROGRAM(X) returns the spectrogram of the signal specified by
    vector X in the matrix S. By default, X is divided into eight segments
    with 50% overlap, each segment is windowed with a Hamming window. The
    number of frequency points used to calculate the discrete Fourier
    transforms is equal to the maximum of 256 or the next power of two
    greater than the length of each segment of X.
 
    If X cannot be divided exactly into eight segments, X will be truncated
    accordingly.
 
    S = SPECTROGRAM(X,WINDOW) when WINDOW is a vector, divides X into
    segments of length equal to the length of WINDOW, and then windows each
    segment with the vector specified in WINDOW.  If WINDOW is an integer,
    X is divided into segments of length equal to that integer value, and a
    Hamming window of equal length is used.  If WINDOW is not specified, the
    default is used.
 
    S = SPECTROGRAM(X,WINDOW,NOVERLAP) NOVERLAP is the number of samples
    each segment of X overlaps. NOVERLAP must be an integer smaller than
    WINDOW if WINDOW is an integer.  NOVERLAP must be an integer smaller
    than the length of WINDOW if WINDOW is a vector.  If NOVERLAP is not
    specified, the default value is used to obtain a 50% overlap.
 
    S = SPECTROGRAM(X,WINDOW,NOVERLAP,NFFT) specifies the number of
    frequency points used to calculate the discrete Fourier transforms.
    If NFFT is not specified, the default NFFT is used. 
 
    S = SPECTROGRAM(X,WINDOW,NOVERLAP,NFFT,Fs) Fs is the sampling frequency
    specified in Hz. If Fs is specified as empty, it defaults to 1 Hz. If 
    it is not specified, normalized frequency is used.
 
    Each column of S contains an estimate of the short-term, time-localized
    frequency content of the signal X.  Time increases across the columns
    of S, from left to right.  Frequency increases down the rows, starting
    at 0.  If X is a length NX complex signal, S is a complex matrix with
    NFFT rows and k = fix((NX-NOVERLAP)/(length(WINDOW)-NOVERLAP)) columns.
    For real X, S has (NFFT/2+1) rows if NFFT is even, and (NFFT+1)/2 rows
    if NFFT is odd.  
 
    [S,F,T] = SPECTROGRAM(...) returns a vector of frequencies F and a
    vector of times T at which the spectrogram is computed. F has length
    equal to the number of rows of S. T has length k (defined above) and
    its value corresponds to the center of each segment.
 
    [S,F,T] = SPECTROGRAM(X,WINDOW,NOVERLAP,F,Fs) where F is a vector of 
    frequencies in Hz (with 2 or more elements) computes the spectrogram at 
    those frequencies using the Goertzel algorithm. The specified 
    frequencies in F are rounded to the nearest DFT bin commensurate with 
    the signal's resolution. 
 
    [S,F,T,P] = SPECTROGRAM(...) P is a matrix representing the Power
    Spectral Density (PSD) of each segment. For real signals, SPECTROGRAM
    returns the one-sided modified periodogram estimate of the PSD of each
    segment; for complex signals and in the case when a vector of
    frequencies is specified, it returns the two-sided PSD.  
 
    SPECTROGRAM(...) with no output arguments plots the PSD estimate for
    each segment on a surface in the current figure. It uses
    SURF(f,t,10*log10(abs(P)) where P is the fourth output argument. A
    trailing input string, FREQLOCATION, controls where MATLAB displays the
    frequency axis. This string can be either 'xaxis' or 'yaxis'.  Setting
    this FREQLOCATION to 'yaxis' displays frequency on the y-axis and time
    on the x-axis.  The default is 'xaxis' which displays the frequency on
    the x-axis. If FREQLOCATION is specified when output arguments are
    requested, it is ignored.
 
    EXAMPLE 1: Display the PSD of each segment of a quadratic chirp.
      t=0:0.001:2;                    % 2 secs @ 1kHz sample rate
      y=chirp(t,100,1,200,'q');       % Start @ 100Hz, cross 200Hz at t=1sec 
      spectrogram(y,128,120,128,1E3); % Display the spectrogram
      title('Quadratic Chirp: start at 100Hz and cross 200Hz at t=1sec');
 
    EXAMPLE 2: Display the PSD of each segment of a linear chirp.
      t=0:0.001:2;                    % 2 secs @ 1kHz sample rate
      x=chirp(t,0,1,150);             % Start @ DC, cross 150Hz at t=1sec 
      F = 0:.1:100;
      [y,f,t,p] = spectrogram(x,256,250,F,1E3,'yaxis'); 
      % NOTE: This is the same as calling SPECTROGRAM with no outputs.
      surf(t,f,10*log10(abs(p)),'EdgeColor','none');   
      axis xy; axis tight; colormap(jet); view(0,90);
      xlabel('Time');
      ylabel('Frequency (Hz)');
 
    See also PERIODOGRAM, PWELCH, SPECTRUM, GOERTZEL.

    Reference page in Help browser
       doc spectrogram

Now from the audio-read function you get the samples Y, and the sampling ferquency fs. Further, you want the temporal resolution of 0.025 s for your spectrum, so that would correpond to something like dt*fs number of samples. Lets just skip the NOVERLAP argument and the NFFT argument aso they will be defaulted. Then you calculate the spectrogram as

NW = round(fs*0.025);
[S,F,T,P] = spectrogram(Y,NW,[],[],fs);
% Then to plot:
subplot(2,1,1),pcolor(T,F,log10(P)),
shading flat,
caxis([-6 0]+max(caxis))
colorbar
subplot(2,1,2)
semilogy(F,P(:,23))
title(sprintf('PSD at time: %f',T(23)))
xlabel('Frequency (Hz)')

HTH

[Yasir Ali]

dear it shows error while I run this line of code

the error is "" Erroe using spectrogram>chkinput X must be a vectoe (either row or column ).

Error in Spectrogram Chkinput (x); ""

what to do to solve that error , may I have your email Id as I can get in touch with you through email please ?