[FFmpeg-devel] [RFC] Channel layouts

madshi dear
Fri Aug 29 16:35:31 CEST 2008

Dominik 'Rathann' Mierzejewski schrieb:
> On Friday, 29 August 2008 at 16:23, madshi wrote:
>> Benjamin Larsson schrieb:
>>> We also need to add a parameter to request the decoded channelmask. For 
>>> example dca have 2 4 channel layouts. Without a requested channelmask it 
>>> would only be possible to get 1 layout.
>> The DTS header specifies which layout the DTS was encoded with.
>> See bitstream field "Audio Channel Arrangement". E.g. value 7
>> means "C + L + R+ S" while value 8 means "L + R+ SL+SR".
>> What sense does it make to request a different decoded channel
>> mask then is stored in the DTS track?
> What if you have 4 speakers but in a different layout than the
> encoded one? With the two layouts above (C+L+R+S encoded and L+R+SL+SR
> speakers) I'd expect the sound to come out of the front speakers only,
> with C+S channels mixed equally with L and R. This could be done by the
> codec or an audio filter if the codec isn't capable.
> Regards,
> R.

As Benjamin replied in the other post I think this is what
a mixer can do. But it should not be the job of a decoder
to do funny mixing stuff. IMHO a decoder should simply
decode and output whatever channel layout, bitdepth etc
is native. Everything that differs from "native" should be
done outside of the decoder, I think. Because that is code
which is not really decoder specific...

Regards, madshi.

More information about the ffmpeg-devel mailing list