Dave Smith Instruments SYNTHESIZERS

I'm pretty excited about the poly-evolver keyboard, but somewhat
concerned as to whether Dave's solution for the low voice count,
polychain mode, is really practical (at least as the normal poly-evolver
implements it).

The poly-evolver manual seems to say that when polychain is engaged, it
completely turns off voice-stealing (it's hard to see how it could do
otherwise, as there's apparently no way to tell the evolver how many
voices can be handled by chained evolvers).

So if you have a poly-evolver or a normal evolver chained to a
poly-evolver keyboard, and play a pad with long envelope release times,
what happens?

On a typical polysynth with voice stealing, the tails of the notes will
get cut off when you run out of polyphony and voices start being stolen,
but this is generally pretty manageable.  If poly-chaining simply sends
new notes along the chain whenever there are no free voices, it seems as
if you'd essentially start _losing notes_ because of the long note
release times -- and this is much less acceptable.

Does anyone have any experience of this?  Does the P.E. handle this in
some clever way...?

Thanks,

-Miles
-- 
=====
(^o^;
(()))
*This is the cute octopus virus, please copy it into your sig so it can spread.

Hmmm, so does nobody really use Evolver poly-chain, or understand how
it works?  That's disheartening...

I guess I can send mail to Dave Smith and ask directly...

Thanks,

-Miles
-- 
Do not taunt Happy Fun Ball.

On Thu, 03 Feb 2005 19:43:49 -0800 (PST), Marzzz@... <Marzzz@...> wrote:

> Hmmm, so does nobody really use Evolver poly-chain, or understand how
>  it works?  That's disheartening...
>  
>  It works...whatcha wanna know??

Er, did you read my original message?

I'll summarize:

I realize poly-chaining works "in general", but my concern is that
will not work very well for sounds like slow-release pads, where you
really need proper voice stealing.

By my reading of the documentation, it sounds like a string of
Evolvers in polychain mode will never steal any voices at all (because
they send any new notes they can't handle to the poly chain output),
leading to missed notes for pads and similar sounds.  A proper
polysynth would steal a voice instead.

It may be that the Evolver handles this case properly; do you know
whether it does?

Here's a test:

  (1) make a pad sound with a very slow release time.

  (2) Now, put your Evolver in polychain mode.

  (3) Play as many notes simultaneously as you have total voices
      (e.g., if you've got two poly-evolvers chained, play 8 notes),
      and immediately release them.  They should continue sounding,
      as the slow release takes a long time to decay.

  (4) Immediately play _one more note_, which is easy to hear even
      with the other voices still sounding, like a high-pitch or something.

  (5) What happens?  Does the new note sound, having stolen a
      voice from one of the still-sounding released notes?  Or do
      you hear nothing, the new note having been sent off the end
      of the poly-chain?

Thanks,

-Miles
-- 
Do not taunt Happy Fun Ball.

On Thu, 3 Feb 2005 23:21:27 -0500, Ravi Ivan Sharma
<sharmalaw1@...> wrote:
> it sends off the poly chain. SImple rule for the 4 voice, 1, 2, 3, 4, 1, 2,
> 3, 4. If you hold down 1 and 2, the other notes are 3, 4, 3, 4, 3, 4 
>   
> if you chain a keyboard and a poly together, you tell the O.S. that you are
> using 8 voices and it goes, 1,2,3, 4, 5, 6, 7, 8, 1, 2, 3 if you then hold
> down 3, 4 and 5, and then play single notes it goes, 6, 7, 8, 1, 2, 6, 7, 8,
> 1, 2, etc. 

Sounds pretty good.  The crucial detail seems to be the ability to
"tell the O.S. you're using 8 voices", which allows it to behave more
cleverly.  I didn't see this mentioned in the polychain section of the
poly-evolver manual -- was it added for the keyboard version?

> This is pretty good. It is not 100 percent perfect. i.e. if the next voice
> to be sounded is 7 because 7 is no longer held, but 7 is still ringing out,
> it will take it. And if 7 is the next in line because 7 is not held down,
> but 7 is still ringing out, and so is 8 and 8 was lifted before 7, it will
> still take 7, not 8. 
>   
> If the rule was the next voice would be the open voice that was released
> prior to any other current open voice, I guess then the logic would be as
> good as it could be. 

Yes, sounds like it (actually that last refinement might be a
noticeable improvement for people with only 4 or 5 voices available.

It's nice to hear that all this sort of detail is being addressed.

Thanks,

-Miles
-- 
Do not taunt Happy Fun Ball.