When exactly is an SSRC RTCRtpContributingSource object updated?

[taylor-b]

Current text:

Each time an RTP packet is received, the RTCRtpContributingSource objects are updated. If the RTP packet contains CSRCs, then the RTCRtpContributingSource objects corresponding to those CSRCs are updated. If the RTP packet contains no CSRCs, then the RTCRtpContributingSource object corresponding to the SSRC is updated.

Is this meant to imply that the RTCRtpContributingSource object corresponding to the SSRC is only updated if a packet contains no CSRCs? I'd suggest having it always updated. There's no extra implementation burden and applications may want to know the SSRC audio level even when CSRCs are present.

[alvestrand]

I think this makes sense.
I'm not sure it makes sense to have the SSRC listed in the contributing sources (it's not, after all, a contributing source), but having the level information consistently available is definitely useful.
(The SSRC doesn't seem otherwise available on an RTPReceiver.)

(Another puzzling question is what happens if we have both simulcast and CSRC. The RTP protocol header layout seems to indicate that CSRC happens per encoding. Is this yet another distraction?)

[taylor-b]

Another puzzling question is what happens if we have both simulcast and CSRC

WebRTC doesn't support receiving simulcast, though. Or am I misunderstanding something?

[aboba]

Another puzzling question is what happens if we have both simulcast and CSRC

I've seen applications that have implemented Opus forwarding instead of mixing for audio conferencing, by forwarding up to N streams (to allow up to N people to speak at once). Each stream goes to its own receiver, with mixing in the browser.

In this scenario (which is not simulcast), the packets may arrive with a CSRC, representing the original source, with the SSRC being one of N allocated by the SFU. Since the packets are forwarded not mixed, and only a single source contributes to each packet, there will be no mixer-client extension. However, it would be useful to have the audiolevel corresponding to the SSRC, which could be used to show the level originating from the CSRC (the original source).