On Thursday 02 November 2006 23:49, Fons Adriaensen wrote: > On Fri, Nov 03, 2006 at 09:23:10AM +1100, Stuart Allie wrote: > > I've just started playing with jack_convolve and I'm having a few > > problems. If I run the program with the default gain factor (1.0) and a > > room-type impulse, the output is horribly clipped - it clips on even a > > small input signal. In order to get a clean output, I have to reduce the > > gain to something like 0.01. Is that normal/expected behaviour? It seems > > odd that the default gain would be 1.0 if a useful value is around > > 0.01... > > It depends on the energy in the IR. If it is a single strong pulse > followed by much lower samples, then a gain of 1 should work. If the > IR contains a lot of high amplitude samples, you will have to reduce > the gain. > > You may want to have a look at JACE : > > http://users.skynet.be/solaris/linuxaudio And i have to admit that i'm not sure that i got the gain thing right in jack_convolve. So, as, contrary to me, Fons knows what he's doing, try his JACE :) I will, too ;) I only wrote jack_convolve because it seemed faster to do to me than understanding/using brutefir's config files (back in the days when brutefir was the only usable convolution engine existing. And it still has millions more features than jack_convolve) ;) Flo -- Palimm Palimm! http://tapas.affenbande.org
