[MPlayer-dev-eng] Another ATSC "a52: CRC check failed" problem

[Mark Haun]

Nico Sabbi wrote:
> the problem is this: usually, when dealing with TS,  AC3 is carried in PES
> packets in private-1 form in a substream (totally idiotic and useless,
> given the nature of TS) beginning with a 0x8n identifier. Certain
> broadcasters do the right thing and remove the substream id, so after the
> pes headers there's only the raw AC3 payload. Unfortunately there's no way
> to know for sure which format is being used, so demux_ts.c probes the
> substream_id in the payload in order to remove the 4 extra bytes (because
> otherwise certain decoders and muxers could be confused) but sometimes
> there's a 0x8n byte in the raw payload (in the second form) that leads the
> demuxer to consider it a substream_id identifier, and consequently to drop
> part of the audio packets.
> 
> If you upload somewhere (maybe mphq2?) a short sample with the whole TS,
> indicating which pid contains the audio track you use, I will try to
> search particular descriptors that may help me to identify which pes
> format is being used.

Hi Nico,

Thanks for that explanation.  Here is a five-second sample from each
channel:

http://www.keteu.org/~haunma/kcpq.ts
http://www.keteu.org/~haunma/ktwb.ts

Regards,

Mark