Re: [fluid-dev] tunnings

[Josh Green]

Hello Viktor,

Ok, I see what you are referring to now.  Sounds like some interesting
work you are doing.  I think I've seen some sort of tuning support in
FluidSynth before, but I'm not sure how complete it is, if at all.  As
for latency, on Ubuntu Linux with the RT Kernel I can get under 2ms with
occasional dropouts and under 4ms with rather rare dropouts.  I'm not
sure what Windows can achieve, but I imagine it would be considerably
worse (possibly not entirely the fault of Windows though, since
FluidSynth only supports directsound).  Latency depends on the buffer
size/count parameters in FluidSynth, but the rest of the system
determines if it is actually usable or not (clicks in audio output).

Best regards,
        Josh

On Sat, 2008-06-14 at 20:55 +0200, Viktor Michna wrote:
> Hello Josh,
> I don't think it's problem of FluidSynth, it does greatly his job and thank 
> you for it.
> Thing is that FluidSynth uses usual equal temperament which sounds slightly 
> dissonant on some intervals (like thirds). This is general problem of whole 
> western music and all conventionally used intruments (piano, guitar, flute, 
> ...).
> 
> http://en.wikipedia.org/wiki/Equal_temperament
> http://www.justintonation.net/whatsound.html
> 
> My research deals with some kind of "cleaning" by small retuning of equal 
> temperament in real time. Maybe if algorithm was ready, it would be good idea 
> to built it into synthesizer, but I'm just at the beginning... There's 
> already method called HMT (Hermode Tuning), but it deals only with some of 
> intervals and more it's probably patented.
> I'm testing on fluidsynth.dll build with patched function 
> fluid_synth_tune_notes() to finally work in realtime (parameter apply).
> BTW: How is it with latency?, I'm using USB MIDI keyboard and there's small 
> delay between the keypress and sound...
> 
> Best regards,
> Viktor
>