I want to somewhat retract my earlier statement on how good the audio quality has been. After being very impressed with the quality on many sessions, this morning I attempted to be (inter)active in a session. I discovered that the delay was intolerably long, over 5 seconds one-way delay between seeing a jabber update and hearing the associated audio. And the jabber typing presumably took some time too, so the real delay was even higher. By the time I heard the chair ask for questions, I typed in a question and the jabber scribe saw it, the next speaker was already talking. It is conventional in speech quality metrics to distinguish between whether the application is one-way or interactive. When interactive, if the delay is too long for the application to properly function, the subjective quality is effectively zero. So, for plenary presentations and sessions where one just wants to listen the audio is great. For sessions where one is used to playing an active role, the quality is completely unacceptable. I captured a few packets and discovered that the audio is MPEG sampled at 22050 over http over tcp. Perhaps we could have a parallel RTSP 8K flow for people desiring it. Y(J)S _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf