Open yanj-github opened 2 months ago
I agree this is a bug. We should be comparing the time of the displayed video (as burnt into the QR code) with the time from the PN audio.
DPCTF:
fixed in version 2.09 as follows
The mediaTime
of the presented audio sample shall match the mediaTime
of the presented audio sample within the tolerance of +40 ms / -120 ms.
I suggest the following: The presented audio sample that has the same mediaTime as the video sample shall be presented jointly within the tolerance of +40 ms / -120 ms.
DPCTF 2024/07/03
Updated according to last comment. integrated into v2.0.10
What are we checking for Audio-Video Synchronization observation?
The spec wording: 9.2.5.4 Audio-Video Synchronization The mediaTime of the presented audio sample shall match the one reported by the video currentTime value within the tolerance of +40 ms / -120 ms.
Is this meant to comparing audio media time with reported application layer current time? I don't think it is correctly checking Audio-Video Synchronization. We are allowing a bigger tolerance for video sample with current time match. "8.2.5.2 The presented sample shall match the one reported by the currentTime value within the tolerance of +/- (1/framerate + 150ms)."
I think we should compare actual presentation time of matching audio and video sample pair.