[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Linphone-developers] Linphone in 32-bit calculations
From: |
Simon Morlat |
Subject: |
Re: [Linphone-developers] Linphone in 32-bit calculations |
Date: |
Thu, 5 Apr 2007 20:01:05 +0200 |
User-agent: |
KMail/1.9.5 |
Actually there shouldn't be a significant difference between the soundcard
clock and the system clock.
Having such big difference makes me think that there is serious problem in one
of them.
Timestamp must reflect the sampling instant of the audio, that the PCM sample
number of the first sample when the packet is decoded.
Sequence number is just monotonic counter, it is used to reorder packets,
detect duplicates and lost packets.
Simon
Le jeudi 5 avril 2007 08:36, Micha Nelissen a écrit :
> Micha Nelissen wrote:
> > I also regularly see:
> >
> > ortp-message-synchronizing timestamp, diff=-1760
> > warning! alsa write underrun!
> > ortp-message-synchronizing timestamp, diff=960
>
> Turns out the cause for this was the system (linux) clock running too
> slowly in comparison with the soundcard providing samples. The following
> piece of code:
>
> if ((diff > delta) || (diff < -(delta * 5))) {
> d->tsoff = curts - packet_ts;
> netts = packet_ts + d->tsoff;
> ms_message("synchronizing timestamp, diff=%i", diff);
> }
>
> Was then executed, which caused a (backward) timestamp jump in the sent
> packets. This in turn causes the receiver to drop packets because they
> are too old.
>
> I think the media streamer should try not to send timestamps that are
> older than the last sent ?
>
> TBH, I don't entirely see the point of the timestamps, since there are
> also sequence numbers already ? Anyone have a good doc on this ?
>
> Thanks,
>
> Micha
>
>
> _______________________________________________
> Linphone-developers mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/linphone-developers