Re: [rtcweb] I-D Action: draft-ietf-rtcweb-audio-02.txt

[Roman Shpount <[e-mail]>]

On Tue, Sep 10, 2013 at 12:26 PM, Jean-Marc Valin <[e-mail]>wrote:

> > To ensure a baseline level of interoperability between WebRTC
> > clients, a minimum set of required codecs are specified below.
> > If other suitable audio codecs are available to the browser to use,
> > it is RECOMMENDED that they are also included in the offer in order
> > to maximize the possibility to establish the session without the need
> > for audio transcoding.
>
> Let's look at the practical effect here, assuming people were to
> actually follow such a recommendation. The only audio codecs that are
> implemented in browsers but not available for WebRTC are the ones used
> for HTML5. According to http://html5test.com/ these codecs are Vorbis,
> MP3, AAC, and uncompressed PCM. These are the codecs that would be
> recommended under such text. None of them is really ideal for
> interactive audio, none of them is implemented in old phones AFAIK, but
> all of them would require time and effort to add to a browser's RTP
> stack. I really don't see the point here.
>
>
My biggest problem with the recommendation text is "other suitable audio
codecs are available to the browser to use". Jean-Marc's comment underlines
the problem. His interpretation is that this refers to codecs already
implemented in the browser. Another interpretation is that this refers to
codecs already implemented in the underlying OS or platform, such as AMR-WB
or H.264. Yet another interpretation implies codecs with no associated
patents, such as G.722 or G.726. Yet another interpretation is that implies
codecs with patent licenses which allows free implementation such as iLBC
or iSAC. What is the intent here? With so many interpretations this
language is completely useless.

I think original intent was to encourage browser implementers to add
support for G.722, AMR and AMR-WB if available in underlying platform, and
H.264, if available in underlying platform or browser. In this regards this
language fails completely. Don't you think it would be better to state
exactly what the intent is and provide the language when implementation of
each of those codecs is not required (meaning if additional license fees
will need to be paid or due to incompatible licensing policy).
_____________
Roman Shpount