On Fri, 12 Sep 2003, Richard Felton wrote:
> I intend to finish my Vorbis document on Monday so I will leave it over the
> weekend to see if there are any more modifications suggested. Although the
> document will be internal I will put a copy of the final version at the
> same address - I know firsthand how annoying it is to find loads of dead
> links when trying to find background info on Vorbis (I know google has
> probably already got a link to my previous posts).
Cool! But there are some things I have to mention ...
[Vorbis encode process]
- The minus within the green circle is a bit
misleading IMO. It should be mentioned that it's
a minus within the "logarithmic-domain".
It's a factorization of the specral-vector into
floor and residue, whereas the floor will be
stored logarithmically.
It's reversed by the dotproduct in a decoder.
I personally really dislike the term
"subtraction" here.
- The meaning of the floor depends on the encoder
strategy. It is currently used as quantization
curve (like the Q-Matrix in JPEG/MPEG), as far
as i could discover. (current = libVorbis 1.0)
[Psychoacoustics]
- "the floor curve is a rough approximation" of what ?
As i said, its meaning depends on the encoder
strategy because of the fact that the factorization
is not well-defined.
(note: <spectral-vector> = <floor> .* <residue>)
( .* = dot-product )
The floor is generally used to scale the residue
vector (like the scalefactors of mp1/2/3,
aac & so on)
Otherwise there would be large demand of many
different code books. ;)
bye,
Sebastian
--- >8 ----
List archives: http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body. No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.