[Fluxus] smoothing bias

[Kassen]

Dear list,

I was playing around with the audio-in stuff (I still think there are things
wrong there) when I ran into a clear bug in the smoothing-bias function.
Consider this;

Welcome to fluxus.
Type (help) for info.
fluxus> (audio-in)
fluxus> (gh 1)
0.11296483129262924
fluxus> (gh 1)
0.09208597987890244
fluxus> (gh 1)
0.06017810478806496
fluxus> (gh 1)
0.07158516347408295
fluxus> (gh 1)
0.11792538315057755
fluxus> (smoothing-bias 0)
fluxus> (gh 1)
0.10764656960964203
fluxus> (gh 1)
0.07180813699960709
fluxus> (gh 1)
0.09894558787345886
fluxus> (gh 1)
0.11659413576126099
fluxus> (smoothing-bias 2)
fluxus> (gh 1)
0.07644085586071014
fluxus> (gh 1)
0.094255730509758
fluxus> (gh 1)
0.0707346573472023
fluxus> (smoothing-bias 1.5)
fluxus> (gh 1)
66436071424.0
fluxus> (gh 1)
5.578410726428508e+18
fluxus> (gh 1)
5.102923371354287e+36
fluxus> (gh 1)
+inf.0
fluxus> (gh 1)
+inf.0
fluxus>

Here "audio-in" is just a utility function I made to set up my audio to
either use the soundcard or Fluxa. The input signal here is just soundcard
input noise. I noticed before that a "bias" of 1 would get the fft stuff
stuck, but here there is actually some clue on what is happening. Right
after I set it to 1.5 the signal explodes and goes to infinity.

I already suspected there was some sort of feedback loop. I noticed that
when running "bars.scm" it reacts properly to some sounds, yet others like a
909 kick that I use will make the rightmost of the 16 bars go up a bit, then
stay there. Playing more such kicks gradually increases that bar and it
won't come back down. To me that indicates some feedback loop.

For reference; I currently have Jack (and Fluxus) set to 2 periods at 44100
Hz with a 1024 buffer, a combination of settings that wasn't possible while
I still had PulseAudio on my system.

Yours,
Kas.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pawfal.org/pipermail/fluxus-pawfal.org/attachments/20110718/dd19767a/attachment-0001.htm>