fix: rebuild lost audio frames instead of dropping them - #8
Merged
Merged
Conversation
- next_pcm decoded whatever arrived and ignored RTP sequence numbers, so a lost packet silently shortened the PCM timeline and left the Opus decoder's prediction state desynced for the frames that followed. - Track the sequence number per audio track and emit frames for every packets. - The frame directly before an arriving packet comes from its in-band FEC copy - anything lost earlier has no copy left, so libopus builds a replacement. - Gaps longer than 10 packets reset the state, which keeps a reordered or wrapped sequence number from generating thousands of frames. - A rebuilt frame is as long as the output buffer, so it is requested at the stream's own frame size rather than the 120 ms maximum.
- The sequence number was tracked as the previous arrival, so a packet arriving after the one that overtook it was decoded a second time: its frame had already gone out, rebuilt from the overtaking packet. The mark also moved backwards, which made every following packet rebuild frames that were already emitted. Four packets on the wire produced six frames. - Treat the sequence number as a high water mark and drop anything older than it. A wrapped 16 bit number reads as behind when the distance is over half the range.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
next_pcm decoded whatever arrived and ignored RTP sequence numbers, so a lost packet silently shortened the PCM timeline and left the Opus decoder's prediction state desynced for the frames that followed.
Track the sequence number per audio track and emit frames for every packets.
The frame directly before an arriving packet comes from its in-band FEC copy
anything lost earlier has no copy left, so libopus builds a replacement.
Gaps longer than 10 packets reset the state, which keeps a reordered or wrapped sequence number from generating thousands of frames.
A rebuilt frame is as long as the output buffer, so it is requested at the stream's own frame size rather than the 120 ms maximum.