[vorbis-dev] RFC draft for Vorbis over RTP

Jack Moffitt jack at xiph.org
Wed May 26 16:17:45 PDT 2004



> This certainly simplifies things in practice with in-band headers 
> or header request over http. Seems a bit of a wash if there's a separate 
> header stream.

Not really.  If you are multicasting, the extra stream doesn't waste
bandwidth unless people are pulling on it afaik.  So it's efficient.  On
unicast, you might as well use a reliable transport since that will be
faster.

So I think the full idea goes something like:

Add a codebook id header to each rtp packet.  This 'id' would be
incremented each time the source restarted if it had to switch
codebooks (or if it can't know, increment always).

Vorbis audio is sent as in the draft, or however in an RTP stream by
itself.

Headers can either be gotten by reliable means (HTTP probably) or from a
separate header rtp stream, on which the headers are broadcast over and
over.  For multicast, the separate RTP stream is recommended.  For
unicast, reliable transport via HTTP is recommended.

Metadata happens in a separate channel.  I'm not sure whether this is
something RTCP should be used for (probably not), or if we should just
make a metadata rtp stream (is there not one of these already?).

> Outside of using a separate metadata stream (which would need to be 
> separate from the header stream, really. otherwise how do you know when 
> it's changed?) I think allowing in-band comment packets anywhere in the 
> rtp stream makes a lot of sense. 
...
> Furthermore, it's much simpler for 
> clients if than having to support a separate access method just to 
> change the current artist/title.

I disagree.  Then do we have comments in the video streams as well?  In
teh speex streams?  Doing this once per stream regardless of all the
differnet media you are sending makes sense.  Only in the trivial case
of single media streams do you get any benefit at all.

I thnk creating such a metadata stream is a pain.  I do hope there is
something already being used for this.  If not, I suggest we write
something ridiculously simple.

jack.
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.



More information about the Vorbis-dev mailing list