[FFmpeg-devel] [PATCH] DCA-in-WAV detection

[Kostya]

On Fri, May 04, 2007 at 02:15:42PM +0100, M?ns Rullg?rd wrote:
> 
> Kostya wrote:
> > On Fri, May 04, 2007 at 11:50:37AM +0200, Michael Niedermayer wrote:
> >> Hi
> >>
> >> On Fri, May 04, 2007 at 09:30:27AM +0100, M?ns Rullg?rd wrote:
> >> > Kostya <kostya.shishkov at gmail.com> writes:
> >> >
> >> > > Here is my attempt on patch discussed on Febryary 2006
> >> > > Thread called "[PATCH] Fix DTS in WAV file playback."
> >> >
> >> > Wouldn't it be simpler to call av_probe_input_format or so on some
> >> > data from the start of the data section of the file?  That would also
> >> > take care of AC3 in WAV masquerading as PCM.
> >>
> >> Note, also see av_find_stream_info() code releated to codec_identified
> >
> > I suppose that this is only WAV related and should be treated there.
> > Also it is gladly is not the common practice so I think checking for
> > those two formats is enough. While I haven't seen AC3 in WAV, DTS in WAV
> > seems to be a common practice (but still a hack).
> 
> I have seen AC3 in WAV done that way.  There is a simple explanation for
> this ugly practice.  Labeling the data as PCM allows one to use stupid CD-R
> software to burn the data as an audio track which can then be played back
> through a hardware DTS/AC3 decoder.  It's not what I call an acceptable
> explanation, but it is an explanation nonetheless.

If you think that is ugly then how can you comment the fact that actual data
stored in WAV is bswap16()'ed (because it will be bswapped back while recording)

While this has no difference for DCA (it accepts any endianness), AC3 parser
(ff_ac3_parse_header) simply refuses to parse anything.
 
> -- 
> M?ns Rullg?rd
> mans at mansr.com
>