regarding music note code can anyone explain this code line by line?

%%
fileName = 'g-piano1.wav';  % File name
[y, Fs] = audioread(fileName);            % Read audio file from input
y = (y(:,1) + y(:,1))*4;                % Decrease 2 channels to 1 (left and right chANNEL CONVERGE INTO TWO channel)
%y = (y(:,1));% extracting first column
y(1:2:end) = 0;   %from big vector take only odd index values         % Do decimation(decreace the no of samples)
frameLength = 4410*2; %samples ratecalculation for two channels
endPart = frameLength*ceil(length(y)/frameLength);  % complete the last frame %ceil is to round off to nearest integer
y(length(y)+1 : endPart) = 0;% setting up in matrix form
f = linspace(1,Fs,frameLength); %creates linearly spaced vector 
%%
harmonics = 0;% SETTING UP haarmonics at zero for beginning the loop structure
for i = 1:round(length(y)/frameLength)      % For each frame
    % Divide audio into frames
    frames(i, 1:frameLength) = y( (frameLength*(i-1)+1):(frameLength*i) )';% diving into two frames
    
    frame = y( (frameLength*(i-1)+1):(frameLength*i) )';same as above
    frame = frame .* hamming(length(frame))';   % Hamming Window
    fframe = abs(fft(frame));                   % FFT
    
    fp = sum(fframe);
    p = sum(abs(frame));
    b = true;
    if(p < 200 || fp < 1000) % Put a threshold for processing
       b = false; 
    end
    
    % Bands
    for i=1:88
        freqBand(i) = mean(fframe( round(bands(i)/(Fs/frameLength) ):round(bands(i+1)/(Fs/frameLength))))^2;
    end
    
    % Plotting
    subplot(3,1,1)
    stem(freqBand)
    subplot(3,1,2)
    plot(fframe)
    xlim([0,500])
    subplot(3,1,3)
    plot(fframe)
    ylim([-1 1])
    
    hold off
    pause(0.1)
    sound(fframe,Fs)
    
    
    % Desicion
    m = find(freqBand == max(freqBand(:)));
    if(b) disp(names(m,:));     % Print the result
    else disp('.'); end
    
    if(b)
        index = 1;
        for i = 1:88
            if(freqBand(i) > 2000)
                harmonics(index) = i;
                index = index+1;
            end 
        end
    
    end
end

[KSSV]

Read about the respective functions used from documentation, this will help you.

[Geetesh Mokhare]

Okay. Thank you But I want a explanation for the code since I am able to understand.

Answer 1

[Walter Roberson]

Yes, I can explain it line by line. My current estimate is that doing so would take approximately 7 one-term university level courses... but as I start mentally reviewing the details, I start remembering more, so it might be 9 one-term courses. Or 10.

Since, for example, fft() does not make any sense without The Calculus, and The Calculus does not make any sense without the theory of real numbers, and the theory of real numbers requires Predicate Calculus, and Predicate Calculus requires a course in Logic and Rhetoric, and those beg for several lectures on comparing the writings of the ancient Greek philosophers...

Answer 2

[Robert Brown]

First trick with understanding this code is that you need the *.wav file. WAV is an audio recording format. Without that, the code won't run.

Since I didn't have the original *.wav file, I went into my computer and found a "replacement wav" file, which was just an audio recording of people talking.

ACTUAL MODIFIED MATLAB CODE IS HERE: (CUT AND PASTE THIS INTO A SCRIPT FILE OR INTO THE COMMAND WINDOW)

%% This is an attempt to answer a MATLAB question submitted to MathWorks
%
% regarding music note code can anyone explain this code line by line?
%       - Geetesh Mokhare
% Solution edits by Robb Brown, 3/29/2021
%

%% Prepare workspace for processing
clear all
close all
clc
format compact
format shortG

%%
% fileName = 'g-piano1.wav';  % Original File name
fileName = '1074_Pos_1_2020_05_06_21_30_23_by_Start_Time_desc.wav';  % Didn't have user file, using this one as as surrogate...

INFO = audioinfo(fileName)   % Get information about the audio file  % Don't end with ; and instead see the INFO

[y, Fs] = audioread(fileName);          % Read audio into array y from input file, and read the sampling rate Fs 

% y = (y(:,1) + y(:,1))*4;              % Decrease 2 channels to 1 (left and right channel converge into  channel)
y = (y(:,1) + y(:,2));                  % Decrease 2 channels to 1 (left and right channel are added together)
                                        % The number 4 appeared to be a gain (4 x louder) 
                                        %    and will also be louder since adding left and right together
                                       

%y = (y(:,1));% extracting first column

% y(1:2:end) = 0;   %from big vector take only odd index values         % Do decimation(decrease the number of samples)
%    This does not take only odd values, it ZEROES OUT every odd value, not the same thing... (and not necessary)

% frameLength = 4410*2; %samples ratecalculation for two channels
%    Don't do this either, make 1 frame be 1 second of data. which is equal to Fs (since Fs is samples PER SECOND)
frameLength = Fs

% endPart = frameLength*ceil(length(y)/frameLength);  % fill up the last frame % ceil is to round off to higher integer
%    Bad logic here, FL * ceil... /FL = ceil...
% here's what you wanted...
endPart = frameLength - mod(length(y),frameLength);

y(length(y)+1 : length(y) + endPart) = 0;% fill in zeros to next full second
f = linspace(1,Fs,frameLength); % now that we've changed to 1 second blocks of audio, the frequency bins are already Hz !!!

%%
harmonics = 0;% SETTING UP harmonics at zero for beginning the loop structure

%for i = 1:round(length(y)/frameLength)      % round wasn't necessary since we added endPart in the code above to make sure it matches 
for i = 1:length(y)/frameLength         % round wasn't necessary since we added endPart in the code above to make sure it matches 

    % Divide audio into frames
    % frames(i, 1:frameLength) = y( (frameLength*(i-1)+1):(frameLength*i) )';% diving into two frames
    frames(1:frameLength,1:length(y)/frameLength) = reshape(y,frameLength,length(y)/frameLength);% dividing into 1-second frames
    
    frame = frames(:,i);   % grab a frame, 1 second of data
    % frame = frame .* hamming(length(frame))';   % Don't do Hamming Window
    frame_fft = abs(fft(frame));                   % FFT
    
%     fp = sum(fframe);  % some sort of attempt at thresholding !?  skip and just display the data as-is
%     p = sum(abs(frame));
%     b = true;
%     if(p < 200 || fp < 1000) % Put a threshold for processing
%        b = false; 
%     end
    
%     % Bands
%     for i=1:88
%         freqBand(i) = mean(fframe( round(bands(i)/(Fs/frameLength) ):round(bands(i+1)/(Fs/frameLength))))^2;
%     end
    
    % Plotting
    subplot(3,1,1)
    plot(real(frame))
    xlabel('Sample number')
    ylabel('Amplitude')
    title('Time Domain Signal')
    subplot(3,1,2)
    plot((1:Fs/4)-1,frame_fft(1:Fs/4))
    xlabel('Frequency Bin (Hz)')
    ylabel('Amplitude')
    title('Frequency Domain Signal')
    % xlim([0,500])
    subplot(3,1,3)
    plot((1:Fs/2)-1,frame_fft(1:Fs/2))
    xlabel('Frequency Bin (Hz)')
    ylabel('Amplitude')
    title('Frequency Domain Signal')
    % ylim([-1 1])
    
    hold off
    % pause(0.1)   % this pauses for 0.1 second, not sure why this is necessary, delete or comment out ...

    'pausing now, press any key to hear this second of audio...'
    pause
    sound(frame,Fs)  % the "sound" command will turn the fframe signal into audio and play through the computer speakers

%     % Decision
%     m = find(freqBand == max(freqBand(:)));
%     if(b) disp(names(m,:));     % Print the result
%     else disp('.'); end
%     
%     if(b)
%         index = 1;
%         for i = 1:88
%             if(freqBand(i) > 2000)
%                 harmonics(index) = i;
%                 index = index+1;
%             end 
%         end
%     
%     end
end

[Walter Roberson]

They asked for "line by line", so you should also have described what

clear all

does. Which would have to include describing workspaces and global variables and persistent variables and Execution Engine and memory allocation schemes (including dynamic memory and the advantages and disadvantages of memory pools), and operating systems, and you should discuss "Cargo Cult Programming"... And discussion of the Execution Engine should get into Turing Machines, and Computability theory, which requires backtracking to Formal Logic.

There is a couple of terms worth of lectures just for that one statement.

Line. By. Line.

If there was even a single line of the code that did not need to be described in detail, then the poster would have narrowed the focus when reminded to do so, but instead they asked again for it to be described line by line. We cannot assume that the poster knows even Grade 1 arithmetic.

[Robert Brown]

The code required interpolations and extrapolations of it's meaning.

In looking back, I am wondering if the original intent may have been to get to a Fast Fourier Transform, or more accurately a Discrete Time Fourier Transform (DTFT), that decodes musical notes into the musical spectrum and identifies notes by letter (A B C D E F G, etc) ?!?! The question and code needed a lot of help, so I did what I could to help along the way, knowing that my answer was not by any means complete or clean. In short, I did what I could to move the user closer to their solution, for what it's worth.

[Robert Brown]

By the way, I learned a lot myself, while trying to figure out that answer... So even if my answer isn't productive for him or anyone else, it was productive for me ! :-)

[Rik]

In case you intend to post more often: please use the tools explained on this page to make your posts more readable.

[Robert Brown]

Thanks Rik, this is my first answer, I'm new, as you could tell