[FFmpeg-devel] [PATCH] Add a parser for DNET (byte-swapped AC3).

Måns Rullgård mans
Wed Mar 2 22:10:21 CET 2011


Reimar D?ffinger <Reimar.Doeffinger at gmx.de> writes:

> On Wed, Mar 02, 2011 at 09:04:54PM +0000, M?ns Rullg?rd wrote:
>> Reimar D?ffinger <Reimar.Doeffinger at gmx.de> writes:
>> 
>> > On Wed, Mar 02, 2011 at 09:23:17PM +0100, Kostya wrote:
>> >> On Wed, Mar 02, 2011 at 08:12:15PM +0000, M?ns Rullg?rd wrote:
>> >> > Reimar D?ffinger <Reimar.Doeffinger at gmx.de> writes:
>> >> > 
>> >> > > Currently we simply byte-swap DNET directly in the Real demuxer,
>> >> > > however this is not generally possible/correct since which two bytes we
>> >> > > need to swap depends on whether the header starts on an even or an odd
>> >> > > byte.
>> >> > > Not useful on its own within FFmpeg, but could e.g. be used together
>> >> > > with a byte-swap bitstream filter or a modified AC3-decoder to implement
>> >> > > more robust DNET support.
>> >> > 
>> >> > Where does the name DNET come from?
>> >> > 
>> >> > > @@ -390,6 +390,7 @@ enum CodecID {
>> >> > >      CODEC_ID_BINKAUDIO_DCT,
>> >> > >      CODEC_ID_AAC_LATM,
>> >> > >      CODEC_ID_QDMC,
>> >> > > +    CODEC_ID_DNET,
>> >> > 
>> >> > I don't think a new codec ID is called for.  Couldn't it be handled the
>> >> > same way as byte-swapped DTS?
>> >> 
>> >> I think it can and should - and there's AC3-in-WAV too to be supported.
>> >
>> > AC3 has only about 17 bits to identify a header on (compared to 32 or
>> > more for DTS/DCA).  Searching for a byte-swapped variant is far too
>> > likely to result in false positives to be acceptable.
>> 
>> Then how did you intend to detect it as ac3 in the first place?
>
> By the fact that the RealMedia header says "DNET"?

AC3 exists outside of realmedia file, you know...

-- 
M?ns Rullg?rd
mans at mansr.com



More information about the ffmpeg-devel mailing list