Hi,
Spectral Band Replication is a bandwidth extension which greatly improves efficiency in AAC today, I was wondering whether it is possible to implement an opensource SBR for Ogg Vorbis.
I recall ones there was an alternative to the Coding Technologies one called "Vplus" or something like that, used along with MP3. I think I has been discontinued why?
Could be possible to create a similar opensource version?

Regards
Enrico

PlusV
http://www.vlsi.fi/plusv/plusvdownload.shtml
http://www.vlsi.fi/plusv/plusv.shtml

PlusV for Ogg Vorbis?
http://lists.xiph.org/pipermail/vorbis-dev...ary/006959.html
http://www.hydrogenaudio.org/forums/index.php?showtopic=153

Vorbis with SBR?
http://lists.xiph.org/pipermail/vorbis-dev...une/012835.html
http://www.hydrogenaudio.org/forums/index.php?showtopic=6307

Thank you. I will check these links

It's not 'discontinued'.. it's just that NOBODY has tried doing anything to improve it (maybe I will.. in like 20 years or so )

Good news.

I do understand from the plusV website that PlusV unlike SBR from Coding Technologies plusV doesn't replicate lower frequencies. The higher frequencies are not generated from the lower spectrum at all, the high frequencies are only sub-sampled to detemine the amount of noise in each subband. On the reconstruction phase, a pseudo-noise is generated according to the auxiliary data. Correct?

Vorbis + SBR = HE-AAC killer ?, very happy testing in the late hours..
http://www.hydrogenaudio.org/forums/index....showtopic=33770

http://www.hydrogenaudio.org/forums/index....topic=34075&hl=

Vorbis has it's own special type of "PlusV", it's called Noise Normalization. The algorithm that's in the source code is a temporary one that can be improved greatly. It's fairly straight forward too and is very pleasent sounding, more so than SBR IMO.

They are totally different things.

Yes, but they are quite similiar in what results they are trying to achieve. I think the problem is that a lot of folks really don't understand the Vorbis source code that well, it's quite difficult to dive into the low-level libraries and routines, but it can't be exactly the same too.

No, no, no, not at all. As Gabriel already said, they are entirely different things. See the analysis by synthesis discussion here; it seems there isn't even agreement about whether the basic principles underlying it's operation are correct.

Dear Garf, can You or Ivan answer on this post?
http://www.hydrogenaudio.org/forums/index....ndpost&p=374008

Thanx!

Where is this discussion again there are so many links being thrown around?. The algorithm seems fairly straightforward to me, I am still trying to understand exactly what it does, but the basic jist of it is that is redistributes the noise energy in specific frequency band that has the lowest SNR. It does this somehow with the residue. I know that it uses an apsort to do so. That's the way I have come to understand and wrote about it in the wiki.

Noise normalisation is there to preserve energy in quantized subbands.
PlusV is a tool to parametrically store and reconstruct higher part of the spectrum.