Thanks for filing the issue!

@Reporter: Could you please share a sample test file/URL which helps us to triage the issue further in a better way. Any further inputs from your end may be helpful.

Mac triage: assigning directly to dalecurtis@ - I know this issue is stale though so maybe it's WontFix?

Hello, unfortunately I don’t have a file or URL to share due to the service that was experiencing this problem is no longer getting any support.

However, from my own investigation and media-internals logs we received from internally reproduced cases this appeared to center around the CHUNK_DEMUXER_ERROR_APPEND_FAILED. With the suberror: “Failed to parse video for decode” (something close to that, iirc).

The filed bug from Matt Wolenetz was what the internal team decided to wait on. It got attached to the MSEBufferByPTS flag. At that point, a content re-encode with proper sample flags was decided to be to much risk/cost as the service was falling out from support.

If you aren’t seeing the issue on a wide-scale then feel free to mark as wontfix.

Otherwise to recreate: package a CMAF segmented MP4 with Widevine encryption, use mp4 trun first_sample_flags to mark first sample as sync. Have multiple GoPs in single track run with sync samples marked as unknown. Expect message in media log about mismatched keyframe analysis from MP4 metadata.

Play content in backgrounded tab ( we never personally reproduced in foreground tab ).

After ~2 hours of playback in this scenario we anecdotally had a high reproduction rate.