[LAD] Selectable limit for polyphony of virtual synth

[Nick Copeland]

Voice allocation really depends on what you want the virtual synth to do.
If you want it to sound like the original then it should use a similar algorithm,
if you want something that sounds better than or like the original then for
something like an Oberheim, it will probably not be the voice allocation that
is lacking on the emulator, more likely the overall quality of the sound that will
prevent the emulator from being as good: whatever you do with voice allocation,
if the emulator does not have Oberheim filters and oscillators it will not sound 
like an Oberheim.

The algorithm was also different between synths, many used a FIFO buffer: if
the voices were all allocated then the oldest voice was stolen. Some would steal
the lowest note pressed, some would steal the highest note. You will eventually
notice which algorithm it uses based on what you are playing - each algorithm
has its defects. Siel did have an allocation algorithm that was supposed to be 
superior in that stealing voices was not that noticible. Never seen the code but 
it apparantly stole a note from somewhere in the middle so the player never lost
either extremes of the scale. I've tried this as well and the results are a noticable
improvement on FIFO.

Pure FIFO stealing is a cheap algorithm however it perhaps does have the most
noticable flaws when being played. VIrtual synths can easily improve on this,
emulators can't, really, and still be something that approaches the original. If
you want an virtual synth to sound like an OB-X the it will have to steal notes in
the same way.

Regards, nick.

"we have to make sure the old choice [Windows] doesn't disappear”.
Jim Wong, president of IT products, Acer




> Date: Tue, 25 Aug 2009 00:13:36 +0200
> From: ralf.mardorf at alice-dsl.net
> To: hannu at opensound.com
> CC: linux-audio-dev at lists.linuxaudio.org
> Subject: Re: [LAD] Selectable limit for polyphony of virtual synth
> 
> Hannu Savolainen wrote:
> > Albert Graef wrote:
> >   
> >> hollunder at gmx.at wrote:
> >>   
> >>     
> >>> One obvious question there is:
> >>> what should the synth do when it reaches the limit?
> >>> There are several things that are possible and afaik implemented in
> >>> synths. It could drop the first note played, or the highest, or ...
> >>>     
> >>>       
> >> Well, that's called voice stealing. Most synths do it, if they don't
> >> have dynamic voice allocation. Usually, you assign voices in a round
> >> robin manner, and the oldest note has to go when you're running out of
> >> voices.
> >>
> >>   
> >>     
> > Ideally the synth should use some kind of priority mechanism when 
> > stealing voices. Killing the oldest one is not the best way. For example 
> > some kind of psychoacoustic algorithm could be used to find voices that 
> > are masked out by the other voices playing at louder levels. Some voices 
> > may have decayed to inaudible levels or their pitch may be close enough 
> > to the new note to be played.
> >
> > Best regards,
> >
> > Hannu
> 
> Hi Hannu :)
> 
> this is a good idea to cover unwanted cutting. "My problem" is, that 
> most virtual synth have enough voices ;). I would like to have the 
> effect of old synth, e.g. listen to Peter Gabrial's pad sounds. It's 
> wanted to hear the cutting, because it has a musical function, it 
> produces ambience. My fault was, that I only referred to the elimination 
> of muddy sound by notes with long release times, I've forgotten to bring 
> up the musical function.
> 
> I like to have the effect that chords will be cut by new cords, similar 
> to the effect for monophonic sounds, with one difference, sometimes one 
> or two notes shouldn't be cut. There are some very good synth with a 
> polyphony of 5 or 6 voices, e.g. the Prophet 5 or the Matrix 6, btw. I'm 
> using a Matrix-1000 (the 1000 is for 1000 sounds in the memory, resp. I 
> didn't check if the battery is still fine, it might be possible that the 
> 200 RAM sounds of my Matrix are lost ;)).
> 
> For most of my external synth and 'virtual' synth I'm missing this 
> effect, they never run out of voices. It's bad not to have enough 
> voices, but if you have enough voices than some charm gets lost.
> 
> I don't know actual Peter Gabriel recordings, but I bet he still uses 
> old synth, e.g. the Fairlight, especially for pad sounds.
> 
> Cheers,
> Ralf
> _______________________________________________
> Linux-audio-dev mailing list
> Linux-audio-dev at lists.linuxaudio.org
> http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev

_________________________________________________________________
With Windows Live, you can organize, edit, and share your photos.
http://www.microsoft.com/middleeast/windows/windowslive/products/photo-gallery-edit.aspx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxaudio.org/pipermail/linux-audio-dev/attachments/20090825/dc926956/attachment.html>