In general what u hear are underrun errors ( where the sound device is not
feed fast enough ).
second generalization is that the linux/sound is not a real-time system - an
active swap can really do ur program, as well as the proper feeding of the
device, "in" . Higher priority tasks can also cause problems, although from
experience, a computationally intensive task at standard priority did not cause
any observable problems.
At one point my IDE drive/controller cause problems ( the time it took to
service the IDE system 'slurred' the sound ). The scsi disks did not cause any
problems.
And Yes, the GC can do u in also. All threads are stopped just so u can do a
GC. The bigger the mem model, the longer it takes. And the longer nothing
happens!
Also in general the sound device buffer is ~64k bytes. That may sound like a
lot, but at CD quality ( ~44khz/channel ) And stereo ( 2 channels ) And 16bit
quality u are talking about ( 44000*2*2) 196000bytes/second. Which, if the pipe
was full, gives u about 1/3 second of sound, before it needs to be feed, or go
into underflow ( silence ).
Not going at full CD speed might help with the underruns.
/gat
Peter Schüller wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hello,
>
> I am wondering; am I the only one who has quality problems with JavaSound
> output? I keep seeing things tested with Linux and so on, but no where does
> anyone mention any quality problems. In all versions of the JDKs I've tried
> it (and that's quite a few; starting with 1.1.8 a while back), sound output
> is always choppy, as if there are regular buffer overruns once a second.
>
> And no, it's not the GC or anything. Plus, in addition the quality itself,
> mins the chopping, tends to be bad aswell. For example, playigng an MP3 with
> JMStudio yields sub-standard quality, while on Windows it's perfect.
>
> I realize sound's probably not a big focus, but since no one seems to be
> mentioning having these problems, I have to ask - is it just me?
----------------------------------------------------------------------
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
