[LAU] Zyn and the art of jack audio/midi

[Atte Andre Jensen]

Fons Adriaensen wrote:

> ISTR (but it's years ago - since the LAC where Zyn was 
> first presented) that Zyn uses FFT based processing for
> some of its algorithms. If this is the case and the FFT
> size has to be some minimal value then there's a problem
> if the jack period is shorter.

I don't know if such a problem is the case with zyn, but when I specify 
the -b 128 everything is just fine. To me it simply seems like a 
user-unfriendly design, where you must specify the buffer size on the 
command line, when theres actually one on possibility, namely the buffer 
size of the running jack server.

I imagine this stems from zyn starting as a non-jack application, 
probably even OSS only, where it makes sense to specify the buffer size 
on the command line. Then I imagine that later when jack support was 
hacked into the thing, the code wasn't really structured in a way to 
make it easy to detect buffersize of a running jack server. This would 
fit with what cal wrote: "I've certainly thought about how to relocate
zyn's application of those parameters until after initialising the 
audio, but I haven't tackled that monster yet".

-- 
Atte

http://atte.dk   http://modlys.dk   http://virb.com/atte