how to cut audio file into small files using silent

hi all

i have an audio file like this

i want to cut it into 5 small files and save them separated to work on. any help

i am using this code but it remove silent only. I don't know how to cut it.

% Read in data and plot it.
[y, Fs] = readwav('F1.wav');
figure, plot(y, 'b-');
%Find the envelope by taking a moving max operation, imdilate.
envelope = imdilate(abs(y), true(1501, 1));
% Plot it.
hold on;
plot(envelope, 'r-', 'LineWidth', 2);
plot(-envelope, 'r-', 'LineWidth', 2);
% Find the quiet parts.
quietParts = envelope < 0.07; % Or whatever value you want.
% Cut out quiet parts and plot.
yEdited = y; % Initialize
yEdited(quietParts) = [];
figure,plot(yEdited, 'b-', 'LineWidth', 2);

Answer 1

[Walter Roberson]

You want to look for transitions from quietParts being true to it being false in order to locate the beginning of non-quiet, and you want to look for transitions from quietParts being false to it being true in order to locate the ending of non-quiet.

A trick involved is to use

   quiet_row = quietParts(:).';   %need it as row vector
   A = strfind(quiet_row, [1 0]);   %silence ends
   B = strfind(quiet_row, [0 1]);   %silence starts

You need to work out the boundary conditions to figure out whether to use those indices exactly as-is or one before or one after.

Also, you need to be concerned about whether the file begins in silence or not, and whether it ends in silence or not. For that purpose it can be useful to pad quiet_row with false or true on both sides (I leave it to you to work out which is appropriate.)

[hanem ellethy]

i did not get it

[Walter Roberson]

Consider for example

quiet_row = [false false false false true true true false false false];

then

strfind(quiet_row, [0 1]) finds the places that non-silence turns into silence, which in this case would be at position 4

strfind(quiet_row, [1 0] finds the places that silence turns into non-silence, which in this case would be at position 7.

The non-silence data to be extracted is then from the beginning to position 4, and from position 7+1 to the end.

If you experiment with [false quiet_row false] and [true quiet_row true] then you will be able to find ways to do this that do not involve having special cases for "might have something at the begining" and "might have something at the end".

[hanem ellethy]

if i use code like this

quietParts = mySignal < someValue;

loudSignal = mySignal(~quietParts);

then how can i save the files separated

[Image Analyst]

With the audiowrite() function.

[Walter Roberson]

loudSignal = mySignal(~quietParts);

is not enough, in that it puts all of the loud parts together in one vector.

The strfind() that I showed gives you starting and ending positions (to within +/- 1, depending how you set up your boundary conditions) so you would loop through those vectors of positions indexing into mySignal to extract a portion to write to a file with audiowrite()

[Joenam Coutinho]

Hello Walter,

I tried to implement this, And I understood the concept. But I am finding it difficult to plot the segments I want.

How would you plot the waveforms that dont include the silent segments.

[Walter Roberson]

   sound_row = ~quietParts(:).';                 %need it as row vector
   starts = strfind([false sound_row], [0 1]);  %sound starts
   stops = strfind([sound_row false], [1 0]);   %sound end
   times = (0:length(sound_row)-1)/Fs;
   for K = 1 : length(starts)
       extracted_time = times(starts(K):stops(K));
       extracted_signal = y(starts(K):stops(K),:);
       plot(extracted_time, extracted_signal);
       %or write the extacted signal to a file if that is what you want
   end

Or, since it is for plotting purposes:

   yc = y;
   yc(quiet_parts,:) = nan;
   times = (0:length(quiet_parts)-1)/Fs;
   plot(times, yc);