[nas] Re: some problems

[Charles Levert]

David Morano <morano at computer.org> writes:

> After playing many more sound clips (beyond those that easily messed up
> before), I have noticed that some (maybe one third to one half) have
> audible dropouts right at the beginning of their play.  This could be
> due to many reason (file startup reading, or some sort of initial
> network delays) but I was wondering if it is at all possible to buffer
> up some amount of sound in NASD **before** starting to play any out to
> the audio device ?  Something like this would probably cleanup the
> remaining dropouts during initial play startup.

I am not sure this is due to starvation.  I get these too and
debugging output indicate no irregularities in timing.  However, I do
sometimes notice some server resets (plural) that I would qualify as
spurious right before the first samples of a run are written.  This
would have to be further investigated.

I don't think there is a programmed buffering delay, but the protocol
uses watermarks in such a way that it asks for more samples at the
beginning than the playing rate alone would provide, effectively
building a buffer.  It think that applications such as mpg123 convert
their buffering delay in terms of buffer capacity and watermarks.


Charles