[vorbis-dev] Spectral band replication

[Stefan T. Böttner]

> Curious.  Along this line of thinking, have you done any checking to see
> how well the higher frequencies encode if you subtract harmonics of the
> lower frequencies first (and add them back in the same ratios when
> decoding)? Actually, it's not that simple, is it.  Bother.  Have you done
> it anyway?
I have done that experimenting with an uncompressed wave file, that I have 
downsampled to 11KHz using sox. I didn't try any coding at all. However, it's 
probably not quite simple to subtract the harmonics before coding, because 
the higher frequencies might have a different phase, so the subtraction could 
actually end up in an "addition" (180 degrees is equivalent to a negative 
sign, ie. we would end up with -2 times this frequency).

Also, when doing it seriously this would break compatibility with existing 
coders. The idea is to use a conventional stream, and then embed a second 
stream (thanks to ogg multiplexing) containing the coefficients of the 
harmonics (ie. their volume relative to the fundamental), which wouldn't be a 
lot in size.

The reason why I'm thinking about all this is because I like Vorbis very 
much, and I wouldn't like this SBR thing to be a reason to use another format.
Not only for me (I can live with 160kbps music files on my harddisk), but I'd 
like to see other people using it too. What should I do if there are no 
artists providing there music in vorbis format?

--- >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 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.