[pulseaudio-discuss] RTP receiver problem, jumpy samplerate

[Erich Boleyn]

[Ed note:  Just scanned pulseaudio list today after a bit of haitus and
 noticed a few messages about RTP issues, comments below]


"Zhang, Vivian" <vivian.zhang at intel.com> wrote:

> Nickie Deuxyeux wrote:
> > Hello everyone,
> > 
> > I am struggling to make my simple RTP setup work correctly. I have one
> > sender and one receiver, both running the latest pulseaudio 0.9.15.
> > Normal "module-native-protocol-tcp" works fine with pulseaudio client
> > (I am using mplayer to test) running from the sender side, but when I
> > start the pulseaudio server on the sender side and start the receiver
> > side, I have the following output:
> 
> I also met the same issue on PA 0.9.15,  :(
> Anyone works fine with rtp module or has some ideas about it?


I use Pulseaudio and RTP in my home for audio through 5 rooms and
7 machines total.

The current RTP code uses resampling based on instantaneous feedback from
the buffer levels and assumptions of the current time.  To work correctly
without continuous audible distortion, it really requires the glitch-free
mode to be enabled and active on a machine with basically no load.  Even
then I have found it is troublesome.

I currently have resampling disabled via a trivial patch, but never
contributed it to the list because the goal is to fix the resampler
control algorithm in RTP.  Not being able to run the same remote audio
on any 2 sets of speakers because they drift apart is really annoying.

I have a bug opened in the pulseaudio.org bugtracker (#477) at:
   http://www.pulseaudio.org/ticket/477

My Wife has let me know it can't persist in this state, and I just got
a bit of free time to rub together, so the next 2 weekends are allocated
for working on this and putting a proper debouncing algorithm into the
resampler control logic in module-rtp-recv.  Please check out my comments
in the bug ticket and/or email me directly if you are interested.

Current estimate is that it will be probably next weekend before I have
a robust fix, but possible that this weekend will do it.

Unfortunately ahead of that is patching the ALSA 1.0.19 drivers to backport
the fix which was recently found for the hda-intel snd_pcm_avail bug(s).
Pulse keeps crashing on the 2 machines with hda-intel sound hardware
right now, also a wife-friendliness problem.  :-P  None of the patches
applied cleanly, so I'm having to reverse-engineer the changes between
1.0.19 and ALSA git-current.  In any case, I'll post that to this list
as well when I get it done.


--
    Erich Stefan Boleyn     <erich at uruk.org>     http://www.uruk.org/
"Reality is truly stranger than fiction; Probably why fiction is so popular"