[FFmpeg-devel] [RFC] SVX8 stereo files and AVCODEC_MAX_AUDIO_FRAME_SIZE

[Stefano Sabatini]

On date Sunday 2011-05-15 16:42:03 +0200, Stefano Sabatini encoded:
> On date Saturday 2011-05-14 22:19:02 +0200, Michael Niedermayer encoded:
> > On Sat, May 14, 2011 at 11:07:24AM +0200, Stefano Sabatini wrote:
> [...]
> > the decoder can just copy the whole into a internal buffer and then
> > return whatever fake partial reads it likes
> > isnt pretty but its simple
> 
> I'm reading the whole audio chunk into the first packet, which is sent
> to the decoder, and which decodes/interleaves the whole audio buffer
> and returns it in little frames.
> 
> Check attached unfinished patches.
> 
> Now my problems is that for 8SVX files audio data format is signed
> 8-bit, which I see is not directly supported as format (why?).
> 
> I could convert samples in the decoder signed->unsigned, but really
> why signed 8-bit is not supported natively (and what would prevent
> from adding it to resample/convert routines?).

Updated work in progress. int8 -> uint8 conversion is done in-codec,
the clipping in delta_decode() looks necessary for avoiding overflow
noise in 8svx_fib.iff (still can't understand why xine doesn't need
it).

Not yet ready for commit (still some minor problems in the decode_frame
codepath), please comment on the overall design.
-- 
FFmpeg = Free Funny Magical Plastic Earthshaking Glue
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0004-iff-8svx-redesing-8SVX-demuxing-and-decoding-for-cor.patch
Type: text/x-diff
Size: 12434 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20110515/84b73f84/attachment.bin>