> Or some _explicit_ feedback from somewhere
downstream the patch
> telling the voice allocator that a particular voice has decayed
> far enough to be a candidate for re-use. My exploratory designs
> for AMS II (gathering dust since four years) did exactly that.
>
>
This "was" a very good mechanism at the times when the first synth were
able to play different sounds for different MIDI channels, but tone
generators were to expensive, because of technical limits,...
Note that the idea is *not* to have such feedback for the reasons
you mention. It is to make such things _explicit_ and user patchable
so a voice allocator can do more sophisticated things than most do
today. For example decide that a new note should not be a new voice
but a continuation of an existing one, depending on some configured
or even patchable conditions.
Ciao,
E.g. as a function of legato or staccato played notes to restart or
continue the envelope?