[LAU] LADPSA or DSSI granular synthesis

[Michal Seta]

On 8/1/07, Chuckk Hubbard <badmuthahubbard at gmail.com> wrote:
> Hi MiS.  You were recently mentioned to me on the Pd list as a fellow
> Pd/Csound user... been having some issues, might contact you off-list...

Sure, I also hang out sometimes on #dataflow channel on
irc.freenode.org (as MiS).  Feel free to write to me.  I also fool
around with untempered tunings :)  I haven't used csound in a  while
(years!) but I have been getting back to it recently.

> I don't know if Csound yet has the ability to act as a DSSI instrument, only
> a host, I thought.  It will act as a LADSPA plugin, for Audacity and Muse
> that I know of (somehow doesn't seem to work with Rosegarden or Ardour for
> me).

in the scons options buildDSSI says "build DSSI/LADSPA host opcodes".
I have not used CSound as a LADSPA/DSSI plugin my understanding is
that it can be both.  Someone correct me if I am wrong.

> But, having played with Csound's granular a good deal, I *highly* recommend
> it for anyone interested in that.  Using a user-defined waveform is
> standard, you can control every aspect, switch between waveforms, do other
> effects on the result, and honestly I think it's capable of things no other
> program is with granular.

I am planning to do a csound rewrite of my (not so) granular pd
performance patch.  It is not quite granular because I rarely use
grains smaller than 500ms but otherwise it is loosely based on some
ideas behind the pulsar synthesis.  But I use only "granular sampling"
technique where I use only live-sampled waveforms (which,
nevertheless, is stored in arrays (aka buffers)).

> Definitely miles ahead of Reason's Maelstrom.

Whatever that is... :)

> Like I said, I don't believe it can be a DSSI instrument yet, but it can be
> controlled by MIDI through aconnect or qjackctl.

and OSC, too.

> I've used both the "fog" and "fof2" commands for great granular effects.  I
> believe "fog" was the one I found most useful.

years ago, I had a simple granulator of soundfiles built with the
oscil opcode.  I had to generate thousands, and even hundreds of
thousands, of lines of score, mind you, but it allowed for exactly the
control over the playback of grains I wanted (every grain was
controlled by a score statement).  No, I did not write the score by
hand.  Today, with kind of CPU power we have, it should be possible to
obtain interesting results in real-time and generating the score
statements with pd (using [csoundapi~], for instance) although I doubt
that pd would be able to provide the kind of resolution one would want
for really cool grain clouds (several thousands of events per second).

./MiS