Re: [fluid-dev] Re: Fluidsynth voice stealing

[Josh Green]

On Fri, 2005-10-14 at 11:56 +0200, Norbert Schnell wrote:
> On 14 Oct 2005, at 9:15, Toby wrote:
> > As already reported, such an algorithm needs to take into account the
> > current position in the volume envelope.  The quietest voice might  
> > be a
> > note that has not started climbing the attack slope yet; we don't want
> > to kill such a note.  The quietest of those voices which are past the
> > attack phase, instead, might be a good candidate.
> 
> BUT (I insist, since Toby put it so clear):
> 
> So this becomes:
>   - Improve volume level rating and limit to notes beyond attack  
> phase (keep track of voice levels?)
> 

I had kind of assumed that, but good to specify for clarity :)

> And needs in addition:
>   - Add rating for notes in attack phase
> 

Do you think there should be an independent rating criteria for attack
phase voices?  I would think it would just be considered max volume for
the volume criteria, and then the other criteria would apply as normal.
If there was one, how would it be calculated?

> Elsewhere:
>    synth.voice-steal.attack-rating
> 
> And what about a little switch for those anyway:
> > default        // The current algorithm (with improvements)
> > oldest        // Oldest voice gets killed
> > quietest    // Quietest voice gets killed
> 

Yeah, I was thinking a way of setting voice stealing presets would be
nice.  So there would be "default", "oldest" and "quietest" presets
which would cause all the criteria factors to get assigned
appropriately.

> Norbert
> 

        Josh

signature.asc
Description: This is a digitally signed message part