Gavan Fantom wrote:
> [a good deal of solid psychoacoustic and harmonic theory]
> ...
> Of course, this is quite hairy stuff to turn into an algorithm, but it
> should at least give you some ideas, and could lead to some heuristics.

Exactly - I think we've got the target bracketed now. ;)  You've shown
what makes voice stealing interesting: there's a nice rich body of
theory, and also a good meaty empirical component of 'what can be
affordable computationally and not sound too goofy in practice?'

I just haven't had, as I said, much opportunity to really absorb the
existing literature on it (not to mention, ahem, patent landscape),
and I'm trying not to become a poster child for Westheimer's Discovery
by rolling something off the top of my head.

I am leaning toward the idea that midisyn should have some stealing
policy a bit more sophisticated than what it has now, plus some way
to mess about with alternative policies, perhaps via LKMs or some
bpf-like hook arrangement. But at the moment I am not doing any work
on midisyn's voice stealing - there are other fish to fry first.

In response to what I think was the OP's question, it will probably
remain true for the near future at least that a low-end consumer
keyboard on a hard MIDI link will be more rewarding for output
than midisyn with an on-board synth.

-Chap