[PD] fast times : [vline~], iem_t3_lib and [block~]

[julien.breval at tremplin-utc.net]

Hello list

I'm working on a patch which consists in a complex interaction between the
following processes :

-> delay lines (with feedback)
-> simple amplitude enveloppes
-> signal mute/unmute 
-> asynchronous triggering timers (non-audio)

These processes should work at very short times (as short as possible). I have
tried lots of different techniques but I still wonder which ones are the best.

These four above processes need two things :
---> (1) control-to-signal conversions with fast interpolation (and no artefact
click)
---> (2) waiting for a very accurate time

Moreover, the delay lines obviously need to include the [block~ 1] object for
achieving very small times (see the "control.blocksize.pd" example file) so I
added it. I don't know if it is really useful to add this object to the other
parts of the patch.


(1) When converting float messages to signals, I used [line~] before, but
[vline~] looks more appropriated for fast changes. There's also [t3_line~] ;
what is the difference between these objects ? Anyway I didn't manage to use
[t3_line~], maybe because it uses much more CPU and I use lots of these
conversions in my patch. [unpack~] from zexy looks good but you can't change its
interpolation time in realtime. So what is the right solution ? I need, for
example, to mute and then unmute a sound during 1 millisecond, so interpolation
time should be about 0.2 or 0.3 ms I think. Maybe it's simply impossible.
Also, the output signal of the enveloppes is generated by a [vline~], should I
use something else ?

(2) Regarding the enveloppes and the timers, I read on the pd-list-archive that
[del] works until 64 samples only (about 1.5 ms). Therefore I replaced all the
[del] objects [t3_delay] in the enveloppes and the timers. Indeed, [t3_delay]
(from IEMlib) is supposed to go below 64 samples. Is it the right method ?


Last question : where can I find some documentation about the IEM_t3_lib ? I
only have the help patches.

Thanks for your help,

-julien