[linux-audio-dev] Realtime convolution - a threading problem?

[Asbjørn Sæbø]

On Wed, Apr 06, 2005 at 06:34:28PM +0200, Sune Mai wrote:

> Convolution is really nice - I use it all the time to get perfect reverb, 
> nice guitar amp-sound etc. The only serious drawback is the high 
> input-output latency of simple FFT based algorithms. [...]

> 2: Another approach is to split the FIR in blocks of different sizes, as 
> can be seen in the bottom figure of this page:
> http://www.music.miami.edu/programs/mue/Research/jvandekieft/jvchapter2.htm
> This approach has an obvious advantage in that it allows "zero" latency, 
> while still retaining some of the computational efficiency of the 
> high-latency FFT based method. [...]

I have not looked into the details of either, but I suspect that this
method is equal to one patented by Lake, and used in their Lake Huron 
audio processor.

Such a method is also described in a paper by Bill Gardner in the
Journal of the Audio Engineering Society.  (The title is something along
the lines of "Efficient convolution without input-output delay".)  I
have heard that Gardner was subsequently informed by Lake that they
already had a patent on this method.

Asbjørn Sæbø