[nas] Re: Sound dropouts using ALSA and NAS on Linux

[Charles Levert]

Jon Trulson <jon at radscan.com> writes:

> 	If possible (I've never worked on the sun code) you might try
> lowering the output sample rate.  Hmm.  would be nice if the sun and hp
> servers also parsed and used the 'outputsection' and 'inputsection'
> options in nasd.conf that voxware uses - would be easier to do these types
> of tweaks.

What I think I will try first, just in case, is to ajust the number of
samples in a chunk according to the sample rate so that the SIGPOLL
rate is constant.  This may be what fragsize does in nasd.conf,
although by looking rapidly at the Linux dda (voxware) code I am not
sure it is actually used.  (The other two frag values seem to be used.)

I.e., 800/8000 is 0.1, while 800/44100 is about 0.018 and maybe the
SunOS audio driver(s) are not able to sustain the latter, but would be
able to sustain 4410/44100.  Sustain is not necessarily the right
word; "being able (or not) to offer this response rate all the time"
might be a better description if the late SIGPOLLs occur because the
kernel is periodically busy on something that prevents it from
responding in 0.018 time.  I'm just speculating here...

If this works, this may be the simplest fix to apply.


Charles