[sdiy] Hardware convolution box?

[cheater00 cheater00]

I was told to look into it after I mentioned reverb and convolution. Let's
see if I can coax more info out of the person who mentioned it.

On Fri, 10 Feb 2017 05:20 , <rsdio at audiobanshee.com> wrote:

> I don't quite understand the "online" part of that. Can you elaborate?
>
> As for zero-delay FFT algorithms, it seems possible to do a moving window
> FFT that updates the output on every sample. The frequency bin values are
> still smeared in time according to the total FFT size, so there is a sort
> of latency in that sense, but at least the output changes constantly rather
> than once per block. The moving window FFT is more expensive that the
> single FFT, of course, but I think there are optimizations that make it not
> as bad as it could be.
>
> Are you sure that this "zero-delay online algorithm for FFTs" is
> appropriate for convolution? My first instinct is that sliding window FFT
> would not quite work for convolution, at least not without some tweaks.
>
> Brian
>
>
> On Feb 9, 2017, at 7:21 PM, cheater00 cheater00 <cheater00 at gmail.com>
> wrote:
> > I was just told there was a zero-delay online algorithm for FFTs which
> would mean 0 delay convolution, but I haven't got a link. Maybe someone
> here heard of it.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20170210/00efdbdc/attachment.htm>