Re: [fluid-dev] warning messages

[Josh Green]

On Mon, 2006-07-10 at 12:48 +0100, cryptoguru wrote:
> 
> Thanks for the advice, will try that tonight ... see what happens!
> 
> Yeah I´m quite excited about this project, 
> getting some fantastic real-time performance from a very tiny box. 
> It´s a case of Dimensions: 65mm(H) x 210mm(D) x 258mm(W) and runs
> nearly silent due 
> to a lap-top type PSU and low-power CPU.
> It'll come with 1GB of RAM, 80GB HDD a 24x4 blue LCD plus a few
> buttons and knobs.
> The idea is for ease of live use and quality of sound.
> 
> 
> I think most synths steal from the oldest sound, but not sure. 
> I much prefer your approach i.e. steal from the quietest sound
> (is that what´s implemented at present? ... will have a dig through
> the source code)
> 
> </stu>

Here is a link to the previous thread on voice stealing.  Perhaps it
might give you some insight into the discussion and perhaps I should
read over it again too ;)

http://lists.gnu.org/archive/html/fluid-dev/2005-10/msg00017.html

Some areas of FluidSynth that I think need work in order for it to be
more useful for live situations:

- Changing some Chorus parameters in real time causes severe audio
artifacts

- Changing LFO frequency causes LFO phase to reset

- Audio gain and clipping.  Currently it can be quite hard to keep the
gain of FluidSynth to a level that prevents clipping.  I suspect that
there might be some clipping internal to the synthesizer itself also (so
even though the output is not maxed, some parameters might cause
internal synthesis stages to clip), but I have not yet verified this.

- Improved voice stealing

Best regards,
        Josh Green