Closed MartinPulec closed 8 months ago
reproducible also simply with -t testcard:mode=Hp24 -s embedded
Thanks for looking into this. Do you think it is fixable? Having to send un-compressed PCM with sync mode is not ideal, but it is better than un-syncronized V/A.
Well, I've slightly ill-designed the new audio compression code while perhaps too much complicating it. I've just simplified it by properly time-stamping the compressed audio.
It is already pushed it but not much tested – just the simple use-cases, where it seem to work. Hopefully I'll get to some more testing in following days.
Forgot to say – --param incompatible
is no longer needed, neither recognized. I'll update wiki accordingly tomorrow.
I've just fixed a small bug (8e8c1f33) that sometimes triggered incorrect behavior (on incommensurable codec and input frame sizes and when audio BPS doesn't much compression, which 4 for AAC, so eg. testcard's 2).
I believe this is now fixed in latest continuous. Tested with both AAC and OPUS.
Thank you.
Originally posted by @alatteri in https://github.com/CESNET/UltraGrid/issues/326#issuecomment-1757020332
Clicks in audio are present if using audio compression with 24 FPS. Steps to reproduce:
Occurs also with OPUS. Doesn't occur if anything of following holds: