On 02/13/2019 03:46 AM, Harald Alvestrand wrote:
> Thanks for confirming the absence of specs. The hackaround that Google
> folks are trying is to send 5.1 as 3 OPUS stereo channels, which doesn't
> make me particularly happy.
No, don't do that. I think the correct solution is to define a new
subtype (e.g. mc-opus) and then just specify an SDP version of the
multi-channel fields of the Ogg header.
For something like 5.1, that would basically mean transmitting the
channel order and coupling. That also means being able to use the
existing libopus surround encoder code, which will have *much* better
quality than explicitly breaking down 5.1 into 3 independent stereo
streams. Even ambisonics should work, though mapping family 3 may result
in a few kB of data in the SDP for the matrices when using 3rd order.
Cheers,
Jean-Marc
>>
>> Cheers,
>>
>> Jean-Marc
>>
>>
>> On 02/12/2019 03:50 PM, Harald Alvestrand wrote:
>>> Hi,
>>>
>>> I wonder if there exists some specification for multichannel OPUS in SDP
>>> - I have searched for it and missed it.
>>>
>>> I note that RFC 8486 (ambisonics) doesn't say anything about SDP
>>> considerations.
>>>
>>> Harald
>>>
>>>
>>> _______________________________________________
>>> codec mailing list
>>> [email protected]
>>> https://www.ietf.org/mailman/listinfo/codec
>>>
>> _______________________________________________
>> codec mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/codec
>
>
> _______________________________________________
> codec mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/codec
>
_______________________________________________
codec mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/codec
