Closed longnguyen2004 closed 8 months ago
Haven't experienced the Unknown (-1)" audio codec
bug, even when streaming for prolonged periods of time (for 100+ viewers who didn't report it happening either).
I don't know if the ssrc will ever change mid stream.
Good work so far 👀
Then it's probably a problem with my client or something else. I managed to catch a failure while stress testing, and asking other people to join and check also didn't reveal anything unusual.
Note about the send interval:
The RFC specifies that
Since implementing them would create unnecessary complexity, and we aren't trying to be standard compliant (we just need it to work), I've came up with estimates for the intervals that would result in ~5 seconds between reports. Feel free to tweak the formula if I'm wrong though.
This should be good enough to be taken out of draft
For other people testing this, please check for:
Unknown (-1)
bug is due to my laptop, or due to me messing up something)My results after 10 hours of continuous streaming:
Unknown (-1)
never happened, so probably something was wrong in my laptop/OS the last timeEdit: Please hold while I test further
Turns out whatever problems I have is due to me using RTSP (note to self: always use SRT). Code is fine and is good to go.
This is still very WIP, missing a few pieces and having some problems that I haven't root caused yet. However it's in a good enough state that other people can test and improve on.
Note: The current implementation timestamps the packet when
sendFrame()
is called. If there are extra processing steps before it that causes the audio and video streams to go out of sync, this will not fix it.TODO:
BaseMediaPacketizer
BaseMediaPacketizer
to prevent "prop drilling"