[Ardour-Users] Good plugins, Re: Hm...

David Kastrup dak at gnu.org
Thu Sep 2 00:45:10 PDT 2010

Arnold Krille <arnold at arnoldarts.de> writes:

> On Wednesday 01 September 2010 14:27:22 Thomas Vecchione wrote:
>> PS I wouldn't use Jamin as a replacement for a channel
>> compressor(Which often can be single band and MUCH cheaper on
>> resources than Jamin is).  It can be a replacement for many things,
>> but the compressors you name I wouldn't use for a mastering compressor
>> either.
> Well, for "simple" compression, the SC* compressors are kind of a standard and 
> perform well. And guess what jamin is using:-)
> But jamin really is overkill when used for a single channel, most single-
> channel signals are mono and jamin introduces a delay of ~10ms which isn't 
> compensated for by ardour as an "insert" doesn't have a chance to have a 
> output port named "latency" to report its intrinsic latency...

Delay management would be an interesting task in itself.  Often for a
plugin, there can be a nonlinear relation between computational
resources (and quality) and latency.  For example, you can implement
large FIR filters much easier using overlap-add FFT techniques, at the
price of latency.

Then there is the situation where something like a dynamic compressor
can work with much less latency (or better quality) if it has additional
access to an unprocessed/rough version of the signal it is supposed to
work with, but with less delay.

And for plugins operating in the frequency domain, it would be nice if
the could negotiate somewhat standardized FFT lengths, overlaps and
analysis/reconstruction windows, and then line up in the frequency
domain instead of doing resynthesis all by themselves.

I have also thought about noise/sound processing boxes in hardware
offering several taps with various delays for live performances.  You
need delay lines anyway for the PA of speakers with distance from the
stage (so that the sound source appears to be the stage), so you could
use the latency for higher quality processing as long as it is more or
less of the constant group delay variety.

David Kastrup

More information about the Ardour-Users mailing list