Bug#540729: ffmpeg: Possible regression - fails to decode AAC

[Reinhard Tartler]

Matthew Wakeling <matthew at wakeling.homeip.net> writes:

> The video recorded by my video camera contains aac audio. This audio is
> correctly decoded by xine, but not by ffmpeg. I have extracted some
> audio and placed it at http://wakeling.homeip.net/~mnw21/sany0118.aac
> for your convenience (56kB only).

I've been able to reproduce the problem. In future, please attach rather
than link example files.

> The bug fixed earlier this year can be found on the ffmpeg bug tracker
> at https://roundup.ffmpeg.org/roundup/ffmpeg/issue800 and was fixed in
> the Fri, 6 Mar 2009 22:36:24 +0000 git commit (no 17860).

FFmpeg is managed via svn, not git. The number you quote (17860) looks
like an svn revision though.

> When I try to decode the above file, I get the following message:
>
> $ ffmpeg -i sany0118.aac -vn -acodec vorbis sany0118.ogg
> FFmpeg version SVN-r19134-4:0.5+svn20090609-1, Copyright (c) 2000-2009
> Fabrice Bellard, et al.

As you can see, debian packages svn revision r19134, which is later than
17860. Anyway, looking at the svn log, I think you mean this change:

------------------------------------------------------------------------
r17856 | alexc | 2009-03-06 20:47:01 +0100 (Fr, 06. Mär 2009) | 3 lines

Fix the channel allocation bug/assumption (issue 800).
Approved by Rob on IRC.

------------------------------------------------------------------------
r17860 | alexc | 2009-03-06 23:36:24 +0100 (Fr, 06. Mär 2009) | 2 lines

If we get an error from ff_aac_parse_header() we should not trust the
header info that it provides.

------------------------------------------------------------------------
r17861 | alexc | 2009-03-06 23:37:21 +0100 (Fr, 06. Mär 2009) | 2 lines

Re-indent after last commit.

------------------------------------------------------------------------

>   configuration: --disable-encoder=h263 --disable-encoder=h263p
>   --disable-encoder=mpeg2video --disable-encoder=mpeg4
>   --disable-encoder=msmpeg4v1 --disable-encoder=msmpeg4v2
>   --disable-encoder=msmpeg4v3 --extra-version=4:0.5+svn20090609-1
>   --prefix=/usr --enable-avfilter --enable-avfilter-lavf
>   --enable-bzlib --enable-libgsm --enable-libopenjpeg
>   --enable-libschroedinger --enable-libspeex --enable-libtheora
>   --enable-libvorbis --enable-pthreads --enable-zlib
>   --disable-stripping --disable-vhook --enable-gpl --enable-postproc
>   --enable-swscale --enable-x11grab --enable-libfaad
>   --enable-libdc1394 --enable-shared --disable-static
>   libavutil     49.15. 0 / 49.15. 0
>   libavcodec    52.20. 0 / 52.20. 0
>   libavformat   52.31. 0 / 52.31. 0
>   libavdevice   52. 1. 0 / 52. 1. 0
>   libavfilter    0. 4. 0 /  0. 4. 0
>   libswscale     0. 7. 1 /  0. 7. 1
>   libpostproc   51. 2. 0 / 51. 2. 0
>   built on Jul  6 2009 06:54:52, gcc: 4.3.3
> Input #0, aac, from 'sany0118.aac':
>   Duration: 00:00:03.52, bitrate: 128 kb/s
>     Stream #0.0: Audio: aac, 48000 Hz, stereo, s16, 129 kb/s
> File 'sany0118.ogg' already exists. Overwrite ? [y/N] y
> Output #0, ogg, to 'sany0118.ogg':
>     Stream #0.0: Audio: vorbis, 48000 Hz, stereo, s16, 64 kb/s
> Stream mapping:
>   Stream #0.0 -> #0.0
> Press [q] to stop encoding
> [aac @ 0x13aed80]channel element 1.1 is not allocated
> Error while decoding stream #0.0
> [this line is repeated lots of times]
> size=       3kB time=10000000000.00 bitrate=   0.0kbits/s    
> video:0kB audio:0kB global headers:3kB muxing overhead 2.825889%
>
> It is possible that the git commit at Fri, 10 Jul 2009 21:53:04 +0000
> (no 19399) has fixed this issue, but I have not checked.
>
> Please let me know if you need any further information.

Well, checking this would be helpful. I've extracted the relevant patch,
you can find it attached.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: aac.patch
Type: text/x-diff
Size: 7525 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-multimedia-maintainers/attachments/20090810/23eff083/attachment.patch>
-------------- next part --------------

I've succeeded to reproduce the problem, I'm currently testbuilding
ffmpeg to see if the patch indeed fixes playback of that file.


...

yes, that patch allows me to listen to your voice "testing 123
testing". I'll include that patch in the next upload.

-- 
Gruesse/greetings,
Reinhard Tartler, KeyID 945348A4