Pitch Detection using AutoCorrelation

I am using auto correlation function to detect pitch of the signal. I am using the equa开发者_如何学编程tion acf(s)=(1/n)*sigma(|x(n)-x(n+s)|)

How do i proceed from here. Should i find the minimum and the pitch is the index of the minimum. In this case the pitch is always 0. Am I doing it correctly? I need help. Thanks.

The first minimum in the autocorrelation function identifies the fundamental period (so logn as the original data is not too noisy). This will be a number of samples, N, so you need to convert this to frequency, e.g.

pitch = Fs / N

where Fs is the sampling frequency.