Re: [codec] Comments on draft-ietf-codec-ambisonics-01

[Jan Skoglund <[e-mail]>]

Hey Jean-Marc,

Thanks for your comments!

On Fri, Feb 17, 2017 at 1:57 PM Jean-Marc Valin <[e-mail]> wrote:

Hi Jan,

Here's some comments based on my review of the ambisonics draft:

3.  Ambisonics With Ogg Opus

>    Ambisonics MAY be encapsulated in the Ogg format by encoding with the

I would suggest replacing the normative MAY by just "can"


Agree


3.1.  Channel Mapping Family 2

>    This channel mapping uses the same channel mapping table format used
>    by channel mapping families 1 and 255.

I would suggest saying "mapping family 1" rather than "mapping families
1 and 255" since otherwise, you have families 2 referencing itself.
Also, we don't know what families 4 and up will look like.


Agree


3.2.  Channel Mapping Family 3

I would suggest removing the "Output Channel Numbering" field because it
is fully equivalent to simply permuting lines of the matrix.


Allowing mixed orders means that there's no unique configuration for a
given total number of channels.
Therefore we need to signal to the decoder which ACN numbers each output
channel corresponds to.



About the matrix -- and as previously discussed privately -- I think it
should probably be stored as 16-bit integers in Q15 format. If any
global scaling is ever required, then the "output gain" field can always
be used.


Sure, this will work.



>    Note that [RFC7845] specifies that the identification header cannot
>    exceed one "page", which is 65,025 octets.  This sets a practical
>    maximum ambisonic order of 10, if full order is utilized and the
>    number of coded streams is the same as the ambisonic order plus the
>    two non-diegetic channels.

This text is unclear about whether partial order 11 is allowed. I have
no strong opinion on the subject, but think it should be clarified.


OK, I don't think we need to to specify a maximum allowed, just point out
the headroom's practical aspect


Cheers,

        Jean-Marc


Cheers,
Jan