FutMan wrote:I made many tests and I think it comes from Audacity. Mkvmerge about if I assemble AAC 6 Channels file that is not spent in Audacity, no problem. Audacity V2.0.2, the file is exported but contains the right voice channel. Audacity v2.0.6, the file will not even be exported. If I export to AAC 2.0 I have no voice at all. 6 Channel AAC files are read by Audacity.
Yes, the only FFmpeg build (version 2.2.2) we could recommend for 2.0.6 uses libvo_aacenc for AAC encoding, and libvo_aacenc is limited to two channels. See
https://trac.ffmpeg.org/wiki/Encode/AAC. But as far as I can tell, if I export to a 2-channel AAC file in 2.0.6, the downmix from six tracks is correct. I don't know why it isn't for you.
Yes on my test, Audacity 2.0.5 produces channel 3 transposed with channel 2 in the exported AAC 6-channel file. For 2.0.5 we recommended FFmpeg 0.6.2 which uses the libfaac AAC encoder rather than libvo_aacenc. I don't know if the channel transposition is a problem with the earlier FFmpeg version or with libfaac.
Most of the AAC encoders have bugs or limitations. So it's a question of choosing which bugs to accept when choosing an AAC encoder to configure in a build of FFmpeg. Some users have been complaining that the libvo_aacenc quality in 2.0.6 is not as good as libfaac for the same file size (and also, you can only specify the quality in 2.0.6 by exporting using "Custom FFmpeg Export" and setting the bit rate there).
Perhaps I will try building FFmpeg with the libfdk_aac encoder, or you can try that yourself if you have the knowledge.
Otherwise perhaps your best solution is to use Audacity 2.0.5
https://code.google.com/p/audacity/down ... .5+windows (because it fixes some other Audacity bugs present in 2.0.2). If the first two channels of the imported AAC file are imported as one stereo track, use the dropdown menu in the name of the first track to "Split Stereo Track". Then open the dropdown menu in the name of the third track and choose "Move Track Up" to move that track to track 2.
Gale