Re: [payload] draft-ramalho-payload-g7110-00 to be discussedinAVTEXT

["Kevin P. Fleming" <[e-mail]>]

On 07/26/2011 06:43 PM, Michael Ramalho (mramalho) wrote:
> Kevin,
>
> Re: "this really isn't much  different from a pair of middleboxes in the
> path of a G.711 call choosing to transcode the audio stream to/from
> G.729 in order to save bandwidth "
>
> I respectfully disagree.
>
> While a transcode from G.711 to G.729 "saves bandwidth" ... it also
> DEGRADES VOICE QUALITY. Thus adjacent signaling elements and/or
> endpoints and/or service providers SHOULD be informed that a transcode
> that degraded voice quality occurred. A natural way to inform the "other
> parties" is via signaling.
>
> A transcode from G.711 to G711.0 "saves bandwidth" (for zero-mean
> acoustic signals) ... but it does NOT degrade voice quality. Indeed,
> after the decompression the IDENTICAL G.711 PAYLOAD is reproduced. Thus
> adjacent signaling elements and/or endpoints and/or service providers
> NEED NOT be informed that it even occurred!
>
> So while I would heartily agree that lossy transcodes be signaled to the
> endpoints, I disagree that they need to be informed of a transformation
> which is truly invisible to the endpoints.

Fair enough. I probably shouldn't have tried to read the draft yesterday 
in the middle of an IETF WG meeting. Mea culpa.

> Do we signal to the endpoints that RTP header compression occurred
> somewhere on the end-to-end path? I don't think so.

No, but there is still signaling between the middleboxes that decide to 
compress/uncompress RTP headers along the path, and the endpoints don't 
even have to be aware that this is a possibility. Granted, this draft 
also allows for such 'endpoint ignorance' if the middleboxes use some 
signaling mechanism (defined elsewhere), but it just feels a bit strange 
to me to have endpoints use a mechanism to indicate to some elements 
that might be in the path how an optional behavior should be employed.

Is it at all realistic to get a static RTP payload number assigned for 
G.711.0? If that was done, then such middleboxes wouldn't need any help 
from the endpoints to select a payload number to use for the transformed 
media.


-- 
Kevin P. Fleming
Digium, Inc. | Director of Software Technologies
Jabber: [e-mail] | SIP: [e-mail] | Skype: kpfleming
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
Check us out at www.digium.com & www.asterisk.org