[FFmpeg-devel] [PATCH] Proper parsing of DTS-HD MA streams - Patch to TS Parser

[madshi]

Aurelien Jacobs schrieb:
> Robert McNamara wrote:
>
>   
>> [...]
>>
>> As we know that E-AC3 is now handled by CODEC_ID_AC3, I have added that to
>> the EAC3 stream handling to that case.
>>     
>
> This is not OK. E-AC3 must be handled by CODEC_ID_EAC3 so that it can be
> remuxed correctly in other containers...
>
> Aurel

In my experience you cannot trust the container information
about whether a track is AC3 or E-AC3. I've seen studio provided
EVO and m2ts files where a track marked as E-AC3 actually
contained AC3 and vice versa.

Also all the DirectShow splitters I know of report E-AC3 tracks
in the same way as AC3 tracks have always been reported.

Furthermore, at least IIRC, MKV does not make a difference
between AC3 and E-AC3, either.

Finally, Blu-Ray main audio E-AC3 tracks actually consist of
an AC3 core with E-AC3 extensions, while Blu-Ray commentary
tracks and HD DVD E-AC3 tracks only consist of pure E-AC3
packets.

How do you want to bring order and logic into this mess?
IMHO it would be best to not differ between AC3 and E-AC3
at all and to have one splitter and decoder for both. But that's
just my opinion, of course.

If you insist on using different CODEC_IDs for AC3 and
E-AC3, then you have to define one more CODEC_ID for
those tracks which contain an AC3 core with E-AC3 extensions.

Regards, madshi.