On Wed, Feb 12, 2014 at 11:11:46AM +0100, Niels Möller wrote: > Kostya Shishkov <kostya.shish...@gmail.com> writes: > > > Well, IIRC you can have some independent channel sets that are alternative > > to > > each other (i.e. different sets of surround channels). So one has to > > determine > > what channel sets you want to decode > > But when parsing the stream, it's not what I want which determines the > number of downmix coefficients present in the stream; that has to be > determined from other in-stream data. Anyway, I found section 8.5.2.1 on > reverse downmixing now, which makes it a little clearer (still not sure > it gives a complete description of all cases, though). So I'll get back > to that later.
The spec is known to be somewhat incorrect and misleading at some places. Just do what you can and fix issues later. > A related question is how all this fits into the libav framework. Can > the application tell libav which channels it really wants (say, 2.0 > stereo or 5.1), and have that information propagated to the decoder, > e.g., dca? Hmm, that's the s->avctx->request_channel_layout? Exactly. And you just stop decoding channel sets if you want less channels. _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel