Reply by Randy Yates January 25, 20072007-01-25
"Robert A." <invalid@invalid.org> writes:

> Also, if I convolve two signals both in the range [-1.0,1.0] is it > normal for the result to be outside that range ?
It is possible depending on what the signals are. Let f1(t) and f2(t) be your two signals. Then f1(t) may in general undergo a maximum magnitude change of \int_{-\infty}^{\infty} |f2(t)| dt. The worst-case "gain" occurs when f1(t) = sgn(f2(t)), where sgn(x) is the sign of x - in other words, when f1(t) has maximum magnitude and sign of f2(t). -- % Randy Yates % "Though you ride on the wheels of tomorrow, %% Fuquay-Varina, NC % you still wander the fields of your %%% 919-577-9882 % sorrow." %%%% <yates@ieee.org> % '21st Century Man', *Time*, ELO http://home.earthlink.net/~yatescr
Reply by Robert A. January 25, 20072007-01-25
Hi guys,

I just added convolution via FFT to my audio program so I can convolve 
two arbitrary signals (sounds), I have a few questions.

In Numerical Recipes in C it says to setup the response in wrap-around 
order, in other tutorials I don't see that being done. Should I be doing 
that ?

Also, if I convolve two signals both in the range [-1.0,1.0] is it 
normal for the result to be outside that range ? Mine always are and I 
have to normalize them, which is fine. But maybe I'm doing something 
wrong and they shouldn't be out of range to begin with ?

Thank you,
Robert A.