[music-dsp] fast convolver

[Steve Harris]

On Tue, Jan 09, 2001 at 01:30:58PM +0100, Wilfried Welti wrote:
> Alexey Menshikov wrote:
> > 
> > Actually i asked sounic foundry about their .sfi format
> > There is no description. as Sounic foundry guy said it just wav encoded,
> > For Acoustic Mirror use simply wav for impulses.
> > 
> > Im seeking for fast convolver too. would be interesting
> 
> For non-realtime usage it could be quite fast using FFT with big blocks.
> for realtime use there is that "lake dsp" patent for using increasing
> block sizes from the beginning to the end of the impulse response. As i heared
> they're a bit restrictive about their patents so who should implement and
> distribute it? Though i guess it would be no problem to make a non-realtime
> implementation just with large blocks, right?

I've been using 512 sample FFTs, seems *realsonably* fast, but my maths
is screwed up somewhere, the results I get out aren't quite right,
there seems to be some smear.

I haven't looked at it in a few weeks, so maybe the bug will be obvious
after at break.

I havn't seen that algorithm, but I'm developing GPL plugins, so patents
are something I want to steer well clear of.

- Steve

dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links
http://shoko.calarts.edu/musicdsp/