Date: Jan 16, 2013 1:59 PM
Author: Derek Goring
Subject: Re: how to determine predominant frequency of any signal by using<br> wavelet toolbox?
On Thursday, January 17, 2013 4:45:09 AM UTC+13, CANER wrote:

> how to determine predominant frequency of any signal by using wavelet toolbox?

Decompose into wavelet details (using wavedec and wrcoef), then take the variance of those details.

The scale corresponding to the largest variance is the dominant timescale.

Frequency is the inverse of timescale, but not the sort of frequency you can plug into a trigonometric function.

But if the signal is stationary (its variance is constant, no matter what the size of the sampling window), you should not be using wavelets for this.