[opus] Channel Mapping Family for Ambisonics

[Timothy B. Terriberry]

Michael Graczyk wrote:
> That makes sense. For now I will focus on encoder only changes. If an
> adaptive pre/post transform had to send side information, would it
> also need to go through the IETF process?

The IANA registry for channel mappings has a policy of "specification 
required", but not "RFC required", so it is possible to specify 
something without going through the full IETF process. Still, if this is 
something you expect to be used outside of Google properties (and it 
sounds like it is), I think going through the IETF process would be a 
good idea (even though that means more work for me).

> Thanks for the link. It looks like that embedding carries metadata.
> Would it be possible to include something like residual coded
> transform coefficients that way?

Yes.

The intent in MPEG TS was to extend this someday to support dynamic 
range control (see, e.g., ISO/IEC 23003-4) and explicit downmixing 
matrices, but no one's done the work for that yet (nor does anyone have 
concrete plans to do it, that I'm aware of). MPEG TS does not have 
global headers (as it is meant for over-the-air broadcast streaming), so 
all of this has to be in-band.

We should probably start keeping track of the space of invalid TOC 
sequences somewhere global so that people don't define conflicting 
extensions.