How to give a variable(here 17658x2 matrix) some proper units desired ?

[y,fs] = audioread('damn-it.wav');
plot(y);  ----

when this code runs audio is stored in 'y' and on plotting ,it shows numbers from zero to 17658 on x-axis. I want to know amplitude at different times so x-axis unit in secs(here 'usec') is desired.

Answer 1

[Walter Roberson]

t = (0:size(y,1)-1)/fs;
plot(t, y)

[Abhishek Maurya]

Thankyou so much.. Can you tell the units in which data has modified into?... And why you chosen 'fs' the number as divisor?

[Abhishek Maurya]

I think i have got my answer. i want you to confirm and elaborate little.... elements in matrix are samples so samples/(samples/sec) gives sec. so unit should be sec.

[Walter Roberson]

Exactly.

[Abhishek Maurya]

I want to calculate the number of times y has become .1. i.e. to solve for t in y(t)=.1 . Please help.

[Walter Roberson]

sum(ismembertol(y, 0.1))

[Abhishek Maurya]

This function just returning zero for every values of y. code:

[yy,fs] =audioread('damn-it.wav');   % Receiving audio
y=yy(:,2);                           % Making it a single channel
sum(ismembertol(y, 0.1));

[Abhishek Maurya]

Thankyou ,i have found my answer.

[i,j]=find(y==.2);
ii=i/fs;  % ans in secs.

[Walter Roberson]

Your line

sum(ismembertol(y, 0.1));

calculates the number of times that y as become 0.1 and then throws away the result of that calculation. You would want to assign the output to a variable.

Your code

[i,j]=find(y==.2);

will rarely find any results because 0.2 is not exactly representable in binary floating point. With the typical 16 bit signed internal representation of wav files, the two closest values that will show up are 0.199981689453125 and 0.20001220703125 which will not compare equal to 0.2 . You need ismembertol() or equivalent to find the locations that are "close enough" to being 0.2 . You might need to provide your own tolerance value for the call.

It would normally be expected that any one value would occur multiple times in a sound. Your code

 ii=i/fs;

would give all of the times at which the value was reached, but would not count them.

[Abhishek Maurya]

Thank you,Yes, you are right. But sum(ismembertol(y, 0.1)); is also not tolerant enough for 16 bit signed representation of wav files. Its giving zero for every values.

[Walter Roberson]

Read the documentation for ismembertol on how to provide your own tolerance