Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 22 views

    For End For End: "Sp01.wav"

    Uploaded by

    sridharchandrasekar
    The document reads in an audio signal, segments it into frames, applies windowing and calculates short-term energy and zero-crossing rate features for each frame. It then separates the voiced frames from unvoiced frames based on these features. The voiced frames are extracted and transformed to the frequency domain using FFT. The magnitude is calculated and converted to log scale. Finally, discrete cosine transform is applied to the log data, and the resulting DCT coefficients are plotted.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd

    For End For End: "Sp01.wav"

    Uploaded by

    sridharchandrasekar
    22 views2 pages
    The document reads in an audio signal, segments it into frames, applies windowing and calculates short-term energy and zero-crossing rate features for each frame. It then separates the voiced frames from unvoiced frames based on these features. The voiced frames are extracted and transformed to the frequency domain using FFT. The magnitude is calculated and converted to log scale. Finally, discrete cosine transform is applied to the log data, and the resulting DCT coefficients are plotted.

    Original Title

    MFCC CODE

    Copyright

    © © All Rights Reserved

    Available Formats

    DOC, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document reads in an audio signal, segments it into frames, applies windowing and calculates short-term energy and zero-crossing rate features for each frame. It then separates the voiced frames from unvoiced frames based on these features. The voiced frames are extracted and transformed to the frequency domain using FFT. The magnitude is calculated and converted to log scale. Finally, discrete cosine transform is applied to the log data, and the resulting DCT coefficients are plotted.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    22 views2 pages

    For End For End: "Sp01.wav"

    Uploaded by

    sridharchandrasekar
    The document reads in an audio signal, segments it into frames, applies windowing and calculates short-term energy and zero-crossing rate features for each frame. It then separates the voiced frames from unvoiced frames based on these features. The voiced frames are extracted and transformed to the frequency domain using FFT. The magnitude is calculated and converted to log scale. Finally, discrete cosine transform is applied to the log data, and the resulting DCT coefficients are plotted.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    of 2

    % % audio input

    [data,fs]=audioread("sp01.wav") % speech signal of 8khz as input,fs-sampling


    frequency
    plot (data)
    frame_duration = 0.010;
    frame_length = frame_duration * fs;% to find the number of samples in each frame
    N = length(data) % total number of samples= 22529
    no_frames = floor(N/frame_length);% no. of frames = 281

    %% framing
    frame=[];
    for i= 1:no_frames
    frame = [frame data(frame_length * (i-1) +1 : frame_length * i)];
    end
    % % windowing
    f_win=[];
    for j=1:no_frames
    f_win= [f_win frame(:,j) .* hamming(length(frame(:,j)))];
    end

    % % short term energy


    ste=[];
    for k=1:no_frames
    ste(k)=(sum(f_win(:,k).^2))/80 ;
    end

    % % zero crossing count


    ZCR1 = [];
    for m=1:no_frames
    x = f_win(:,m);
    ZCC = 0;
    for m1=1:length(x)-1
    if ((x(m1) < 0) && (x(m1 + 1) > 0 ))
    ZCC = ZCC + 1;
    elseif ((x(m1) > 0) && (x(m1 + 1) < 0))
    ZCC = ZCC + 1;
    end
    end
    ZCR1 = [ZCR1 ZCC];
    end

    % %seperating voiced frames using STE and ZCC


    voiced=[];
    for h=1:no_frames

    if (ZCR1(h)<30 && ste(h)>0.000132)


    voiced(:,h)=frame(:,h);
    else
    voiced =[voiced zeros(80,1)]

    end
    voicedpart = voiced(:);
    end
    % SEPEARTING FRAMES OF VOICED PART
    voice=[];
    for i= 1:no_frames
    voice = [voice voicedpart(frame_length * (i-1) +1 : frame_length * i)];
    end
    % REMOVING ZEROS FRAMES
    voiced_sep=voice( :, all(voice,1) );
    % fft
    FFT=[];
    for s=1:length(voiced_sep)
    FFT=[FFT fft(voiced_sep(:,s))];
    mag=abs( FFT);% magnitude
    phase=angle(FFT);
    end
    % logarithm
    LOG=[];
    for s=1:length(mag)
    LOG=[LOG log(mag(:,s))];
    end

    % DCT of log data


    voice_dct=[]
    for s=1:length(LOG)
    voice_dct=[voice_dct dct(LOG(:,s))]
    end

    figure
    plot(voice_dct);

    You might also like