Re: [iiwusynth-devel] iiwusynth

[Josh Green]

On Thu, 2002-05-09 at 17:10, Peter Hanappe wrote:

<cut>

> 
> The problem, as far as I see, is the long release phase of notes.
> The MIDI files I played (I mainly use iiwuplay) have notes of an
> average length of 200 msec. However, the release phase easily takes
> 800 msec. That means, if you're only playing 10 notes at once,
> there are 50 voices being played by the synthesizer. It also uses
> 80% of the time to synthesize nearly inaudible sounds.
> 
> Now, the synthesizer cuts of sounds at -100 dB. I just changed it to
> -80 dB, and there's already a gain. I'm not sure what to do about it.
> We could have an option like "--faster-release".
> 
> Maybe I made a mistake in the calculation of the release time?
> 
> 

I think when max polyphony is reached voices should start being killed
using certain criteria. This is how the OSS AWE sound font driver does
things. Voices could be rated by whether they are currently in volume
envelope release stage and what level it is at. The voice with the
lowest priority would be dropped.

It would also be neat to have dynamic load detection / voice allocation
routine. If the user could set a max amount of CPU usage iiwusynth
should consume. Then voices could be dropped not based on a fixed amount
but based on the load. I guess the tricky part here would be in coming
up with a nice way to calculate load.

> 
> It's a lot of work to create a good soundfont. I would like to bundle
> a small and free soundfont with the iiwusynth. Any suggestions anyone?
> 
> 

Yeah, I get frustrated every time I try and make a sound font with
Smurf. Its not even really possible yet with Swami (still need a paste
routine). I've got a lot of ideas for making things easier, such as auto
FFT based sample tuning, batch operations (loading of multiple samples,
setting multiple generators at the same time), python scripting, batch
GUI graph editing of generator parameters with the ability to select
arbitrary controls from different presets/instruments, etc. 

Unfortunately I'm already rather overwhelmed with just getting Swami to
basic functionality. I'm confident that the new architecture is really
flexible, but sometimes I feel like I'm hacking a bit when it comes to
putting it all together. I'm happy to see the amount of support that
iiwusynth is getting from other developers. I just wish Swami had the
same, of course I have yet released a tarball of the beast. I guess DSP
stuff is where all the fun is :) Anyways. I'll quit ranting for now.
Cheers!
        Josh Green