26 Jan
2005
26 Jan
'05
1:25 p.m.
Hi,
onsdagen den 26 januari 2005 16.13 skrev Mark Constable:
--
http://spamatica.se/music/
Steve Harris wrote:
Some years ago I experiemented quite a lot with brutefir and the result can be
extremely good, though very CPU consuming... And if I understand things
correctly, unless Lake DSP's patent can be circumvented or proven invalid
(Oh, I think it's already been proven invalid, but not in a legal
sense...IANAL) we will have a hard time getting the CPU-cost down.
Still brutefir is a working OSS solution that plugs into jack, no guis
though ;-).
If anyone wants to experiment there's always http://noisevault.com/ with
plenty of impulse responses.
I've got some config files for brutefir also, if anyone is interested. It's
kind of tricky to get going though...
Regards,
Robert
Well, I'm not a reverb expert, but my
impression is that convolved
reverbs are the way to go*. The actual code is relativly simple, and we
allready have open source examples, and its possible for people with
decent recording equimpent and access to spaces with interesting
acoustics, or artificial reverb units to capture thier own impulses. They
can also be synthsised using a method similar to raytracing.
Sounds encouraging. Surely there is a really obvious and
common set of a dozen or so presets that would cover 90%
of most users needs ?
To me, the main need for a highly configurable interface
is because there is no obious "Hall 1" thing to click on
so, sure, we need reverbs interfaces with lots of intricate
and barely understandable options... if these mythical presets
were well targetted and convincing then most folks would not
need to understand the inner sanctum of audio engineering
to have some half decent onboard reverb.
The downside is that the parameters you can
control are very limited, and
it burns quite a lot of CPU power, and there can be latency. The upside
is that the sound can be very good.
Most encouraging. There is an obivious technique to prevent any
latency problems, but bits
of it are probably patented by Lake DSP, though the situation is a bit
murky.
If I had more time I'd like to take the code in brutefir and make it into
a DSSI plugin with a simple dropdown select-an-impulse, set gain type
interface.
Sounds like a plan.
I have a sizeable library of impulses, but
unfortuantly I dont know the
provenance of many of them.
Uhm, what's that mean in English :-)
* In the interests of full disclose, I have an
addiction to convolution
that should be decalred :)
Sounds like a safe enough addiction... Steve, what would you
outline as the most direct path to solving the general problem
of not having decent open sourced reverb code under linux ?
In my wildest dreams I'd like to think "we the community"
could get it together to even fund an effort like this to
ensure "we" have a solution sooner than later. I mean, it's
going to cost me, and anyone else, quite some dollars to
buy decent hardware so it makes practical sense to me to put
those dollars into an open sourced solution that is then
available to everyone. Pay once, use everywhere.
--markc