[vorbis-dev] Channel coupling in Vorbis

[Jean-Marc Valin]

> If I understand properly, your objection is that directional perception
> varies with frequency, and the ambisonic channels don't take this into
> account. Can't we leave that to the codec? I don't see why it shouldn't be
> free to shift things in and out of the W channel, or is it something more
> sophisticated you're worrying about?

I think there is some processing that can be done before the codec and that can
do stuff the codec can't do. A simple (though not necessarly common) example
would be an audio cut for which the left channel is the same as the right
channel, but delayed by a small amount of time, or (more likely) filtered with a
certain transfer function. In this example, it would be possible (and relativly
simple) to extract (and quantize) the transfer function between the channels and
only encode the difference between the prediction and the actual channel, which
can be made very small. 

        Jean-Marc 


-- 
Jean-Marc Valin
Universite de Sherbrooke - Genie Electrique
valj01 at gel.usherb.ca

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/