1276238 - Firefox can't play Opus tracks without CodecDelay

Status: RESOLVED WONTFIX

(Core :: Audio/Video: Playback, defect)

Description

[kagami]

Attachment: with_delay.webm

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:46.0) Gecko/20100101 Firefox/46.0
Build ID: 20160425114621

Steps to reproduce:

Latest FFmpeg from git (any commit since April, 10) produces Opus WebM files without CodecDelay field on remuxing. It was reported here: https://trac.ffmpeg.org/ticket/5509 however no actions has been made yet.

Commands to generate example files:
$ ffmpeg -f lavfi -i testsrc -f lavfi -i sine -t 5 -c:v vp8 -c:a opus with_delay.webm
$ mkvinfo with_delay.webm | grep delay
|  + Codec delay: 6.500ms (6500000ns)
$ ffmpeg -i with_delay.webm -c copy without_delay.webm
$ mkvinfo without_delay.webm | grep delay
<nothing>


Actual results:

First WebM can be played without any issues.
Second WebM can't be played in Firefox: "Video can't be played because the file is corrupt.".
Both WebMs can be played in desktop standalone players (like mpv) perfectly fine, Chrome/Chromium also play them without any issues.


Expected results:

I'm not sure why exactly Firefox strictly requires CodecDelay field but since other players handle files without CodecDelay well, it should probably be relaxed in Firefox too?

I've only found this bug: https://bugzilla.mozilla.org/show_bug.cgi?id=951770
>In particular, we should reject files whose CodecDelay track property doesn't match the value from the OpusHeader CodecPrivate
Is that the reason such files recognized as corruped?

Comment 1

[kagami]

Attachment: without_delay.webm

Comment 2

[Jean-Yves Avenard [:jya]]

:derf, the file is rejected because the codec delay set in the webm do not match the preskip value.

Should this check be relaxed? and if so, should we even care about the codec delay setting?

Comment 3

[Timothy B. Terriberry (:derf)]

Echoing what I said in #opus on Friday:

<+derf> jya: I'd say if the file has a non-zero preskip value and no CodecDelay, then we're well within our rights to reject it.
<+derf> Any codec-agnostic demuxer (to the extent such a thing is even possible) is going to do the wrong thing with that file.

Comment 4

[Jean-Yves Avenard [:jya]]

The file is broken, the muxer should be fixed.