[Tftb-help] RE: Tftb-help Digest, Vol 4, Issue 3

[Eric Chassande-Mottin]

> Regarding your reply, I have one question: energy is defined as : 
> energy=average_power x time = mean[abs(sig).^2] x [length(sig)/fs], but 
> refer to the program, energy = mean[abs(sig).^2]. I'm wondering how do 
> you consider the part of [length(sig)/fs] when implementing this 
> sigmerge.m.From your reply, you seems assumed fs=1, then how about 
> length(sig) portion? and if test on real-life signal which fs is not a 
> unit sample, eg,fs=8khz. this program still can work on such real-life 
> signal?

sigmerge make sure that the ratio of the energies of signal and noise 
is equal to a given number. both signal and noise are assumed to be of
same duration (=length/fs). the factor length/fs cancels when
doing the ratio.

note that this definition of the SNR is ONE possibility among others
and that the definition used in your field may not match this one.

eric.