On Sun, Jan 07, 2007 at 01:27:11PM +0100, Atte André Jensen wrote:
> 5) Anyone (other than me) looked inside the code and interested in
I worked on AMS together with its principal author, Matthias Nagorni.
My contributions were some modules, a complete rewrite of the signal
flow mechanism (wich was originally much more complicated without
any benefit), and the ALSA and JACK audio interfaces.
At that time Matthias was part of the group working on audio at SuSE.
He has now moved on to project management and has no longer the time
to work on AMS.
The place of AMS has largely been taken by Om, and a integration of
the best of AMS into Om would IMHO be a interesting development.
Unfortunately Om does not really support the logarithmic 'control
voltage' concept that is at the heart of AMS and of the analog synths
it was designed to emulate. The reason being AFAICS that Om's author
seems to dislike it and prefers linear controls - we had some heated
exchanges about this. Mixing both paradigms into a single program
*could* indeed confuse some users, but that is a IMHO a matter of
documentation and education.
The quirks of AMS are related to the MIDI interfaces and voice
assignment. While all the other processing code of AMS is fully
dynamic and reconfigurable at any time, these functions use static
objects that are hardwired together, e.g. voice assignment relies
on a hidden feedback channel from the ENV generators to the MIDI
Adding multi-timbral patches and dynamic polyphony would require
a complete rewrite of these parts, and considerable changes to
how things are seen by the user, for example the feedback path
mentioned above would need to become explicit, the flexible
assignment of MIDI controllers to module parameters would become
more complicated, etc. The main reasons why it was never done were
that it would be a major rewrite, and break the large collection
of examle patches that Matthias had created.
It's still on my to-do list, but not for any time soon...
Lascia la spina, cogli la rosa.