Re: [rtcweb] JSEP: Issues with a=ssrc and RTP payload type switching

[Magnus Westerlund <[e-mail]>]

Harald Alvestrand skrev den 2015-06-08 17:17:
> Den 08. juni 2015 17:15, skrev Simon Perreault:
>> Le 2015-06-08 09:09, Harald Alvestrand a écrit :
>>> This has another implication:
>>>
>>> When doing PT switching for comfort noise, you MUST have a CN with the
>>> same clock rate as your normal audio codec.
>>>
>>> CN is exactly the type of PT-switching that, if it occurs at all, occurs
>>> very frequently.
>>
>> I'm not following. We're discussing generating a new SSRC on clock
>> rate-switching, not on PT-switching. For the CN example, you would
>> typically include in your SDP one CN payload type per possible clock
>> type so that you can switch payload types without switching the clock rate.
>
> Yep. It's only because I've seen SDP trying to match CN/8000 with
> OPUS/48000 that I'm mentioning it - it's obvious, but only after it's
> bitten you.
>

Yes, this is likely one of these things that the ones that considered 
the issues see as obvious, but is in fact not until you actually tried 
it and therefore could have benefited from a note about the need for 
having one PT configuration per timestamp rate used by the codecs that 
may use the CN.

As the RTP usage is in IESG review, I am hesitant to add such discussion 
into the RTP usage document. As it is really a configuration issue, I 
think it can be discussed in JSEP, but the audio draft might be an even 
better fit adding to the CN text.

Cheers

Magnus Westerlund

----------------------------------------------------------------------
Services, Media and Network features, Ericsson Research EAB/TXM
----------------------------------------------------------------------
Ericsson AB                 | Phone  [phone]
Färögatan 6                 | Mobile [phone]
SE-164 80 Stockholm, Sweden | mailto: [e-mail]
----------------------------------------------------------------------