Re: [rtcweb] Call for Consensus Regarding Selecting Recommended Audio Codecs

[Tim Panton <[e-mail]>]

On 11 Jan 2013, at 12:33, <[e-mail]> <[e-mail]> wrote:

> It is clear from the discussions on the rtcweb e-mail reflector that the only additional codecs to be considered are limited to the following very small subset of 3 codecs: AMR, AMR-WB and G.722. 
> In addition to G.711, these 3 codecs cover almost all legacy devices dedicated to voice services. They are consequently needed and sufficient to be supported by WebRTC to make it an attractive and future proof technology for usage in all environments including mobile and for interoperability use cases with most of all legacy voice terminals.
> 
> AMR and AMR-WB are indeed the most widely supported voice codecs in hundreds of millions of legacy mobile devices.
> G.722 is royalty free (including a Packet Loss Concealment solution provided in ITU-T Software Tool Library) and is the codec used for HD Voice / DECT-Cat IQ fixed devices
> 
> Furthermore, considering that the reason for excluding AMR and AMR-WB from WebRTC was the licensing issue, there is no reason to NOT support AMR-WB and AMR at WebRTC level if these codecs are already implemented on the device.
> 
> Therefore I support option 1 and propose the following specification according this:
> AMR-WB, AMR and G.722 are RECOMMENDED TO IMPLEMENT by WebRTC end-points
> AMR and AMR-WB MUST BE supported at WebRTC level by WebRTC end-points on 3GPP mobile devices already implementing AMR and AMR-WB (*)
> 
> (*) note that the way these codecs are supported at RTC Web level is left open to implementors: either by a WebRTC specific software implementation of these codecs or by using APIs to access hardward implementation. 


I'm against this sort of statement in general, it only serves to fragment the webRTC landscape.

If we do end up making this sort of statement it needs to say:

"AMR and AMR-WB MUST BE supported at WebRTC level by WebRTC end-points on 3GPP mobile devices already implementing AMR and AMR-WB provided no additional license or agreements are needed for their use in this way."

The point being that just because the codec is in the silicon, it does not mean it is licensed for just any use, the license may be limited by channel count or direction. (I've seen phones where AMR decode is permitted but not encode).

I strongly encourage anyone who wants to mandate g722 to make a call over 722 from their laptop in starbucks before dragging rtcWEB back a decade or two.


Tim.