# counting the number of frequency in wav file

3 views (last 30 days)
faiz hazizi on 24 Apr 2018
Commented: faiz hazizi on 29 Apr 2018
hi, i would like to count the number of frequency on my sample audio file. it is a recorded drum audio file which contain 8 beat of kick pedal in 7 seconds. so the problem is, how to make a simulation to count the number of kick pedal in this audio file (.wav)
here are my .wav file
t=linspace(0,length(y)/fs,length(y));
plot (t,y)
faiz hazizi on 25 Apr 2018
sorry sir, but i dont understand ? why is it need to calculate the rms value ? and what is the relationship between rms value and findpeaks ? i do found the coding of the findpeaks which is numel(findpeaks(your_signal)) but it didn't works on my work. and i don't understand what is the "(your_signal)"should be ? is it .wav file or what ?
Jan on 26 Apr 2018
@faiz hazizi: Remember that a diagram without labels does not explain anything. What do the two diagram represent?
The original signal has positive and negative values. You are not interested in the maximum value, but in the maximum amplitude. This can be positive or negative. Using abs helps to move all values to the positive side. If you have a single signal (mono sound), this is exactly the same as RMS: The mean of a scalar is the scalar, and Root of Squared value makes the sign positive. For a stereo signal, RMS considers both channels.
There is no "relationship" between RMS and findpeaks. The first one converts the signal to do what you want, the second one finds the peaks.
"your_signal" is the "y" in your code: The variable in which the audio signal is stored.

Jan on 26 Apr 2018
yRMS = sqrt(sum(y .^ 2, 2)); % Root mean square, same as abs(y) for mono signal
[pk, loc] = findpeaks(yRMS, 'MinPeakHeight', std(yRMS));
See doc findpeaks for more methods.
faiz hazizi on 29 Apr 2018
thank you sir, appreciate that, i'll try run the program first. if there's is any in problem regarding to this program i'll ask again sir, thank you

Von Duesenberg on 25 Apr 2018
A possible workflow:
%compute the envelope of y (your signal)
%assuming it's mono, and adjust the second
%parameter of the envelope function with
%successive plots of myEnv
[myEnv, ~] = envelope(y, 3000, 'peak');
[pk, loc] = findpeaks(myEnv, 'MinPeakProminence', .5);
%get the number of peaks
nbPk = length(pk);
Von Duesenberg on 26 Apr 2018
Edited: Von Duesenberg on 26 Apr 2018
If you do
size(y)
And the second output says 2, then your audio is stereo; if it says 1, your audio is mono.
Jan on 26 Apr 2018
@faiz: "Mono" means, that the sound has been recorded with 1 channel. "Stereo" uses 2 channels, and needs 2 microphones. Ask WikiPedia for details.
If a sound is recorded with 2 channels, the output of audioread must have two columns also. See: doc audioread.