[xiph-rtp] Updated Vorbis and new Theora I-D's

[Michael Smith]

On Fri, 04 Feb 2005 09:34:31 +0100, Tor-Einar Jarnbjo
<Tor-Einar at jarnbjo.de> wrote:
> Ralph Giles wrote:
> 
> >You can't simply skip the first octet, there is other important data in
> >the other bits. You could skip it for header packets, but you can only
> >skip the LSb for audio packets. In any case it sounds like consensus is
> >the VDT field should just be removed entirely.
> >
> >
> Sure, I forgot about that. Why about using the two VDT bits for the
> packet count field instead, allowing 63 and not just 15 Vorbis packets
> to be bundled in one RTP packet. Are there any probable scenarios where
> 63 Vorbis packets would not exceed the MTU limit?

Yes, but you generally don't want that many vorbis packets in a single
rtp packet.

Digital silence (which isn't terribly uncommon, though it'll usually -
one would think - be only short fragments) would typically fit many
more than 63 packets into a typical MTU (I think it ends up being 1
byte per packet. Maybe it's 2?).

The downside is that you then have a very large gap between packets -
particularly in live situations, this risks starving client buffers.
Existing vorbis-in-ogg-over-http streaming often runs into this
problem.

I doubt allowing a large number of vorbis packets in an rtp packet
would have any significant uses.

Mike