Re: [rtcweb] Conclusion statement for Recommended Audio Codecs call

[Roman Shpount <[e-mail]>]

On Thu, Feb 21, 2013 at 10:20 PM, Ralph Giles <[e-mail]> wrote:

> On 21 February 2013 17:29, Cullen Jennings (fluffy) <[e-mail]>
> wrote:
>
> > I've never understood why G.722 was disabled in the builds. Does anyone
> have any insight into that?
>
> As discussed at IETF 84, we believe the pair of G.711 and Opus are the
> appropriate MTI choice to ensure interoperability.
>

Can you explain why iSAC is included? It is a much more complex codec then
G.722. It is not standardized or reviewed by any standards body. Its IPR
situation is unclear, since there is no mechanism to claim IPR against it.
It would also require a substantial amount of maintenance. I would even
say, that in presence of Opus it has no value, except  to connect to one of
a couple instant messenger networks.


> Experience with the web's long-lived historical content has made me
> conservative about data formats proliferation, since historical
> formats must often be supported indefinitely. I am therefore in no
> hurry to support additional codecs in our webrtc implementation.


I understand why this argument makes sense for MTI codec, but as far as
additional codecs implemented in the browser it does not hold water. There
is a big difference between image or audio formats and codec support. If
image format is not supported some portion of the web page will not be
displayed. If optional codec is no longer supported, end points will
negotiate one of the MTI codecs, and in the worst case we will end up with
G.711 call. It will not sound as good but it will still work. Right now
there is a good reason to support G.722 taking into account number of end
points for which it is the only wide band codec supported. In the future,
when this situation changes, browsers can drop this codec, and whoever is
still left over using G.722 will have to switch or use G.711.Not
implementing it now and thus not giving the time for the market to catch up
is unreasonable.

BTW, during the time when MTI codecs were discussed, one of the arguments
not to include G.722 (or other codecs in MTI list) was that WebRTC
implementers would do a reasonable thing and support a comprehensive list
of codecs to insure widest possible interoperability. Currently this is not
the case.

Finally,even though supporting codecs is not free, amount of effort spent
arguing about G.722 on this list probably exceeds resources necessary to
maintain G.722 for the next few years.
_____________
Roman Shpount