[vorbis] Spectrum duplication (2)

[Christian.Buchner]

> Yes, this is of course a problem. But can the sheer _concept_ of 
> regenerating the higher end spectrum from the lower end with some 
> auxilary data be patented? Well.. I don't know how it works, but 
> software patens in general suck in my oppinion :-(.

It appears to have a similarity to "motion vectors" in video coding.
Use a vector, point it to another part of the video frame and take
the data from there. ;)

For audio, you would have to add an attenuation to compensate for
the fact that the higher frequencies usually have much lower energy.
And if you want to make it really fancy, then you could use a
polynomial term to provide a variable attenuation over frequency.
Also you would have to think about how to code all this auxilliary
data in a very efficient way.

I've previously played with applying the concept of "motion vectors"
to audio (both along the temporal axis as well as along the
frequency axis). I got pretty good results coding the power
spectrum. But the phase was too random to profit from this coding
trick (I used an FFT based codec, not a MDCT). I may have produced
the slowest encoding audio codec ever, but haven't applied for the
Guinnes Book of Records yet.

Christian

--- >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-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.