Re: Re: [fluid-dev] FS 1.1.2 alpha sound rendering issue

[Bernd Casper]

Hi David,

 

you hit the point exactly. After Sven compiled the latest svn, all the crackling has gone completely, as confirmed from both the other testers who had the crackles too. Also memory usage and latency seem much improved. So much thanks for the changes. So far, it seems not to have something to do with cpu-cores.

 

I take the liberty to forward their answers to the fluid-dev list. Great work, Gents.

Many thanks again

Bernd.

 

****************************

 

This works great! The single-note crackling problem is fixed!. I am
also able to reduce the buffer size X count to 512 X 3 with no sound
breakup! This gives very good latency as well.

CLW

 

****************************

 

  Hello Bernd

Thank you for distributing this. I have just tested it and didn't want
to stop playing. The touch performance and the tonal clarity are just
sensational. I took buffers down to 4 and suspect I could go lower.
This plays like an electronic organ there is no delay. I was not able
to detect any polyphony limit or drop out even on heavy registrations
with couplers and many notes held. I had polyphony set at 1024 on each
of 3x instances of Fluidsynth.

Regards
Rick

 

****************************

 

----- Folgende Nachricht wurde empfangen -----

Absender: David Henningsson

Empfänger: fluid-dev

Zeit: 2010-09-12, 22:14:27

Betreff: Re: [fluid-dev] FS 1.1.2 alpha sound rendering issue

2010-09-12 18:32, Bernd Casper skrev:
> Hi David,
>
> some days ago I reported about strange crackling effects using single waveforms. With the new jOrgan release (containing FS 1.1.2 final) the effect appears on other machines, too. Please have a look at the thread in the jOrgan user list:
> http://jorgan.999862.n4.nabble.com/jOrgan-3-11-first-beta-td2535918.html#a2535918

Right, so I fixed ticket 86 and I've done a small optimization (svn
revision 369) by not starting threads if they are not likely to be used.
That optimization could have as side effect that this problem is now
solved, but I'm not certain as I don't really understand how it appears.

Could you check if it is fixed in the latest svn, and if it is not,
report whether the synth.cpu-cores option has any effect on whether the
bug appears or not?

Thanks,
  // David