[MPlayer-dev-eng] Re: [PATCH] channel reordering for 6ch audio

[Corey Hickey]

Frank Aurich wrote:
> Alexander 'Lazy Ranma' Ponyatikh wrote:
>> Mplayer's support for 6 ch sound is crappy. The only thing that works
>> correctly is AC3 audio (at least with -ao dsound). Every codec
>> produces or requests channels in its own order, and mplayer almost
>> never reorders them. As result channels are often mixed up.
>>
>> I made some research on channel ordering of various codecs and
>> drivers, and made some patches to make them correctly reorder channels
>> to/from mplayer default order. I took L R SL SR C LFE sequence as
>> mplayer default, since it's used by ad_liba52.c, af_hrtf.c,
>> ao_dsound.c.
>>
>> ao_dsound.c
>> Native order:   L R C  LFE SL SR (windows default)
>> Expected input: L R SL SR  C LFE (mplayer default)
>> Driver correctly reorders channels
>>
>> ao_pcm.c
>> Native order: L R C LFE SL SR (windows default)
>> Expected input: native
>> Reordering required
>>
>> ad_faad.c
>> Native order: C L R  SL SR LFE
>> Output order: native
>> Reordering required
>>
>> ad_liba52.c
>> Native order: L R SL SR C LFE converted from planar LFE L C R SL SR
>> (using MMX)
>> Output order: L R SL SR C LFE (mplayer default)
>> Codec correctly reoders channels
>>
>> ad_libvorbis.c
>> Native order: unknown
>> Output order: native
>> I didn't find any standards for Vorbis, b/c many codecs differently
>> reorder channels. Even files encoded/decoded by the same tools got
>> their channel order completely mixed up.
>>
>> ad_pcm.c
>> Native order: L R C LFE SL SR (windows default)
>> Output order: native
>> Reordering required
>>
>> ae_lavc.c
>> Native order: L C R  SL SR LFE
>> Expected input: native
>> Reordering required
>>
>> ae_pcm.c
>> Native order: L R C LFE SL SR (windows default)
>> Output order: native
>> Reordering required
>>
> Corey Hickey addressed the problem of wrong channel order in AAC/faad 
> audio a while ago:
> 
> http://thread.gmane.org/gmane.comp.video.mplayer.devel/32141/focus=32141
> 
> I've been using this patch since then, and it works as expected. I can't 
> say which one of the patches is better, you decide.

Since I don't have time to work on my approach and finish the patch, I 
have no objection to anyone making other methods work.

-Corey