rtpbench conformance cases: ava

date 2026-10-21T15:52:00+00:00, image rtpbench:fe41c0856bb3, commit aivc-b95f5af-wave-16-g687f998, runs 10, kernel 6.8.0-147-generic, python 3.14.8

ava, clean

netem per direction: none

True value = measured − error; the error lies within:

• [-17.0, +17.0] ms: caller's labelled end; ±5 dB of label threshold, median (calibrate --speech)
• [-0.4, +0.4] ms: capture times; wire timing max abs error, clean (calibrate)

Wall-clock steps during the run: none.

RTP pacing, ava

The agent's RTP as the caller received it, ms, median [min, max] over calls: late, p99 of how much later than their timestamps say the packets arrived, against the earliest before them: the jitter buffer a receiver needs to play 99% of them in time; jitter, the RFC 3550 interarrival jitter estimate, its median over each call. On a clean network both are the agent's own pacing (and a carrier's); a profile adds its own.

Pauses: where the agent stopped sending RTP for a second or more and went on with the same stream, outside a hold (silence suppression, RFC 3551 4.1), median [min, max]: per call; how far the RTP clock fell behind across each, ms (the pause less what the timestamps advanced: 0 if they kept time, as RFC 3550 5.1 requires); and how many packets after one carry the marker bit, as 4.1 asks. A carrier that hangs up on RTP inactivity may hang up on these.

Calls

invite-retransmit · clean · run 1 ✓ pass

refer-lost · clean · run 1 ✓ pass

bye-lost · clean · run 1 ✓ pass