[CELT-dev] RTP payload

Jean-Marc Valin jean-marc.valin at usherbrooke.ca
Sat Nov 22 13:51:33 PST 2008

Timothy B. Terriberry a écrit :
> How does the decoder know how many frames are in the packet? I see an
> nframes SDP parameter, but it appears to be optional, and while there
> are timestamps, they are only for the beginning of the packet, requiring
> a subsequent one to determine the number of frames. What am I missing?

My idea was that 1 frame/packet would be the default since that's what
most people are likely to use (packing multiple frames defeats the
very-low-delay purpose of CELT). More than one frame would still be
allowed, but it would have to be signalled somehow. I guess there's
always the option of spending a bit to flag whether there are more than
one frame in the packet, but I feel like that bit would be mostly wasted.


More information about the celt-dev mailing list