On Monday 11 July 2011, Andrew Suffield wrote: > On Mon, Jul 11, 2011 at 10:08:58AM +0200, David Henningsson wrote: > > I agree that it feels strange, but question is if it matters? > > Yes, that is the question. I can't really tell; there's too much going > on here that I don't fully understand yet. > > > For > > release volumes, things should always ramp down anyway so any value > > >= 1 should be okay. > > After looking more closely, I note that if we enter an envelope > section and the current value is out of range, it will immediately > skip to the next section. That also seems wrong and may interact > poorly with the quirky max values. Maybe the right thing is to clip to > the new range when entering a new section? Unclear if this is ever a > real issue.
I've spotted a practical issue related to envelopes, and I think that may be related to your observations. The scenario is as follows. Compile the current SVN for x86 arch. with debug turned on and using floats (it is the default setting for auto-tools, use -Denable-floats=yes for CMake). Run fluidsynth limiting the polyphony to 32 and load the VintageDreamsWaves-v2.sf2 soundfont (available at hammersound.net). Example: $ fluidsynth -osynth.polyphony=32 VintageDreamsWaves-v2.sf2 The first program is named "FM Bells 1" and has a long release. Play with this sound some quick arpeggios until the max. polyphony is exceeded. You will get some messages at the console like these: fluidsynth: debug: Polyphony exceeded, trying to kill a voice fluidsynth: debug: Killing voice 56, index 15, chan 0, key 67 Now, change the program to 5, corresponding to the sound "El Cheapo Organ" which has short release, with no decay. > prog 0 5 > channels chan 0, El Cheapo Organ Playing slow legato scales with it shows that many notes sound with the wrong envelope (no sustain), and also when there is sustained sound sometimes the sound is distorted, garbled. Regards, Pedro _______________________________________________ fluid-dev mailing list fluid-dev@nongnu.org https://lists.nongnu.org/mailman/listinfo/fluid-dev
