Re: [fluid-dev] SoundFont parameter analysis

[Garth Hjelte]

Thanks for responding!

>The spec says: "Generators at the Preset Level are instead considered 
>"relative" and additive to all the default or instrument level generators 
>within the Preset Zone."
>Both instruments set modEnvAttack and modEnvRelease to 100 ms. The preset 
>generator of e.g. Polysynth2 sets them to approx. 70 for attack and approx. 50 
>for release.
>
>This makes an attack of 100 ms * 70 = 7 sec
>and release of 100 ms * 50 = 5 sec.

I know you didn't write the spec =) but doesn't the "relative" and "additive" 
mean:

Attack  of 100 ms + 70ms = 170ms
Release of 100 ms + 50ms = 150ms

The spec says additive, not multiplicative. Unless I'm missing something?

>The modulation in this case affects the filters cut-off freqency. Thus it 
>needs 7 seconds for the filter to fully open. This is pretty much what I hear. 
>Dont know about the 1.5 sec you refer to.

Just what I heard Vienna do, and I consider that the gold standard since Emu 
created it at the time SoundFont was designed. But 7sec is closer sounding that 
60ms =)

>Note however that what you hear is strictly speaking not correct. fluidsynth 
>currently processes the modEnvAttack in a linear way. According to the spec 
>(8.1.2 #26) it should however be convex. See here for further reading:
>https://github.com/FluidSynth/fluidsynth/issues/153

Thanks for the reference, interesting reading. 

Garth Hjelte
Sampler User