Re: [SoX-devel] New code for CMT

[Richard F. <rf0...@bl...>]

Hi there - sounds like there's some good stuff happening here. Sox is a
great app that I've used over many years - it would be great to make it
and its effects more LADSPA-friendly!

I'd be inclined to put the plugins into a NEW library rather than CMT
though, because CMT is not actively maintained at the moment and it uses
a specific C++ approach that I suspect is unnecessary for your purposes
(LADSPA itself is a C interface). I don't mind helping a little to get
folk off the ground - though the interface is fairly well documented and
there's a guide or two on the web.

Channel data in LADSPA is handled by use of multiple ports - each
individual channel is always mono. This can be a little restrictive - in
general I guess that SOX would have to unpack interleaved audio for
LADSPA processing and then pack it up again before forwarding it on. (So
yes, a vol plugin would work if you didn't unpack the data, but this is
a special case - in general, the results you'd get would be bad.)

BTW, have a look at Steve Harris' set of LADSPA plugins
(http://plugin.org.uk/).

Have fun,

--Richard


-----Original Message-----
From: Reuben Thomas [mailto:rr...@sc...] 
Sent: 28 May 2007 11:12
To: ri...@mu...
Cc: SoX developers
Subject: New code for CMT


[SoX devs: this is an email to the inventor of LADSPA and maintainer of 
CMT. I think it would be great if we could push some of our effects into

CMT. I'll email the list separately about SoX-specific issues.]

I'm a maintainer of SoX (http://sox.sf.net/) and I've just added LADSPA 
support to SoX, so that LADSPA plugins can be used as SoX effects (if
they 
have one input and one output, for the moment).

I'm also keen to push effect code out of SoX into LADSPA plugins where
that 
is appropriate.

Looking at the list of plugins in CMT, and the list of desired plugins,
I 
seem to have a good fit. There may be some SoX plugins I can simply
throw 
away in favour of existing CMT plugins; some others with extra
functionality 
that can be added to CMT plugins; and some definitely implement desired 
plugins (in the CMT tasks list).

I was wondering if you have any time to help. In particular, I don't
know 
C++, and the existing code is all in C. It's all GPL-compatible.

Whether you can help with coding or not, it would be interesting to
discuss 
the best approach to getting plugins into CMT, and also which plugins
SoX 
can simply discard in favour of CMT plugins. Here's the list of SoX
effects 
which I believe could be made into LADSPA plugins:

biquads
chorus
dcshift
dither
earwax
echo
echos
flanger
mcompand
phaser
reverb
stat
swap
synth, tremolo, vibro
vol

I already have one specific question: the SoX "vol" effect can cope with
any 
number of channels, but SoX ensures that the channel data is interleaved
in 
the input, one sample from each channel at a time. Would I be right in 
thinking that the amp_mono plugin in CMT can cope with this, since 
(presumably) it treats each sample individually?

Let me know if you're interested. I'm currently putting a lot of effort
into 
opening SoX up. Until now most of the work has been on the formats and
I/O 
side, adding support for libraries like libsndfile, ffmpeg and libao.
I'm 
keen to let SoX benefit from other libraries, and let other libraries 
benefit from it, and in the process reduce unnecessary duplication in
the 
open source sound processing community. I posted to the Linux audio dev
list 
recently asking about LADSPA; my conclusion was that SoX can't use
LADSPA 
for all its effects, so that for some applications its existing API is
still 
useful, and should be solidified and promulgated. So I'm not against 
diversity either. But I am keen to free up open source programmers to
create 
new tools and improve existing ones, rather than keeping on reinventing
the 
wheel.

-- 
http://rrt.sc3d.org/ | Troy is in the Iliad, not in the sand (A. N.
Wilson)