[Matroska-users] "ERR0B1: SimpleBlock at 5979 track #2 is not a keyframe" error thrown by mkvalidator tool
slhomme at matroska.org
Wed Feb 14 08:10:37 CET 2018
2018-02-13 17:01 GMT+01:00 Moritz Bunkus <moritz at bunkus.org>:
>> Indeed, the audio frame are not marked as keyframe. That is incorrect
>> for audio tracks (unless some codec has some special requirement, but
>> that's currently not the case).
> Uhm, in general you're not entirely correct. I know of at least two audio
> codecs that actually do differentiate between key and non-key frames: one
> of the older RealAudio variants and the very-much-current TrueHD. TrueHD
> calls them "sync frames" vs "normal frames". Only "sync frames" contain
> full headers, and decoders can only start decoding from "sync frames" but
> not from "normal" ones. Hence, "sync frames" are flagged as key frames
> whereas "normal" ones aren't.
> Opus doesn't belong to that category, true, but mkvalidator must not assume
> that all audio frames have to be marked as key frames.
True, I didn't refine to that level. Which makes me think, the codec
description in the Matroska specs should also state which codec uses
> That file has a totally different problem, though: the audio track doesn't
> have a CodecPrivate element… Therefore its data cannot be decoded. mkvmerge
> does therefore not recognize the track, and even trying to ffmpeg to decode
> it fails (VLC produces… something awful, too).
VLC is using libavformat for muxing Matroska. So it should give the
same (bad) result as ffmpeg. It may have different version from the
CLI version though.
> Kind regards,
> Matroska-users mailing list
> Matroska-users at lists.matroska.org
> Read Matroska-Users on GMane: http://dir.gmane.org/gmane.comp.multimedia.matroska.user
Matroska association Chairman
More information about the Matroska-users