sorting data into excel file

Hey,

as a part of constructing a calssifier a have built a algorythm which calculates featrues over vocal WAV samples,

a file is loaded, than framed and normalaized, than go through procsses ocer features as- SHIMMER , JITTER etc.

I want to arrange each resault from each vocal sample into a excel table by a loop.

for instance, if i use MFCC 16 , i would like the raw that explains 'X' vocal signal sample to have 16 colums ,each one will reperst a value of the standard deviation sice the result will be a matrix. for the JITTER result i would like a different colum a so on for each feature.

the loop would run through all the vocal signal samples (100 for a start) and organize each feature calcultion over each sample into a diffierent colum.

since i'm not so strong at MATLAB i would love some help figuring this out.

thanks!

[Mathieu NOE]

hello

have you started to write a code so far ?

[Omer Aroetty]

I have written a code that calculates all features i need on a single vocal signal. Now i want to understand how to organise the results in excel as i asked above

Answer 1

[Pratyush Swain]

Hi Omer,

In order to store feature extraction for audio files in a excel file , you can create to table to store the features and use "writetable" function in MATLAB to store the data in excel. Please refer to an example workflow as follows:

numFiles = 1 % Modify it as per your number of files

numFiles = 1

% Preallocate a table to hold the feature values, please note, 
% I have used 17 columns - 1 for Index(ID), 14 for MFCC , 1 for Jitter
resultsTable = table('Size', [numFiles, 16], ...
    'VariableTypes', repmat({'double'}, 1, 16), ...
    'VariableNames', ['ID', strcat('MFCC_SD_', string(1:14)), 'Jitter']); 

% Iterate over all your files
for i = 1:numFiles
    % Read the WAV file
    % Replace it with your file names,using a demo file in this case
    [audioIn, fs] = audioread("Counting-16-44p1-mono-15secs.wav"); 
    
    % Perform Preprocessing of audio(framing, normalization,etc)
    % ---------------------------------------------
    
    % Feature Extraction
    [coeffs, ~] = mfcc(audioIn, fs); % mfcc function in MATLAB returns coefficients
    [rows,cols] = size(coeffs);
    fprintf("Size of coefficients is as follows --> rows: %d and cols: %d ",rows, cols)
    mfccSD = std(coeffs); % Standard deviation across columns 
    
    % Modify/Replace with your jitter calculations
    jitterValue = rand(1) ;
    
    % Fill in the table
    resultsTable.ID(i) = i;
    resultsTable{i, 2:15} = mfccSD;
    resultsTable.Jitter(i) = jitterValue;
end

Size of coefficients is as follows --> rows: 1551 and cols: 14 

disp(resultsTable)

    ID    MFCC_SD_1    MFCC_SD_2    MFCC_SD_3    MFCC_SD_4    MFCC_SD_5    MFCC_SD_6    MFCC_SD_7    MFCC_SD_8    MFCC_SD_9    MFCC_SD_10    MFCC_SD_11    MFCC_SD_12    MFCC_SD_13    MFCC_SD_14    Jitter 
    __    _________    _________    _________    _________    _________    _________    _________    _________    _________    __________    __________    __________    __________    __________    _______

    1      1.8194       3.3625       1.2825       0.6205       0.64335      0.41481      0.3109       0.31299      0.2583       0.24075       0.26204       0.20797       0.26111       0.20545      0.91808

% Export the table to an Excel file
writetable(resultsTable, 'VocalFeatures.xlsx');

You can modify the above example workflow as per your usecase/implementation. You can also modify table structure and add other columns for MFCC coefficients,shimmer,etc at the start of table declaration.

Also, please refer to following resources:

1- Table - https://www.mathworks.com/help/matlab/ref/table.html

2- writetable - https://www.mathworks.com/help/matlab/ref/writetable.html

Hope this helps.