Vorbis dev - Aug 2000 - Why LSP?

(Disclaimer: this is not an LPC vs. LSP question)

After looking at the Vorbis code I was wondering why you were using LSP to
quantize the spectral envelope instead of simply quantizing the cepstrum
(DCT(log(envelope))) or modified cepstrum (DCT(envelope.^alpha)). To me it seems
like when the information is already in the frequency domain, there's no
need to
go back to LPC. Also, I think a DCT can be faster that a Levinson-Durbin
recursion. This is why I'd like to know whether there's a special reason
I'm not
aware of for quantizing the LSP.

        Jean-Marc


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

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

> (Disclaimer: this is not an LPC vs. LSP question)
> 
> After looking at the Vorbis code I was wondering why you were using LSP to
> quantize the spectral envelope instead of simply quantizing the cepstrum
> (DCT(log(envelope))) or modified cepstrum (DCT(envelope.^alpha)). To me it
seems
> like when the information is already in the frequency domain, there's
no need to
> go back to LPC. Also, I think a DCT can be faster that a Levinson-Durbin
> recursion. This is why I'd like to know whether there's a special
reason I'm not
> aware of for quantizing the LSP.
Well, you're correct that cepstrum be implementing a very similar idea. In 
fact I'm encoding the LSP(log(envelope)) so in a way it's a closer
parallel
than is obvious ;-)  The reason I'm not quantizing the DCT is that LSP 
quantization does not produce ripple and irregularity throughout the envelope 
like quantizing a DCT would (and the DCT quantization error would tend to be 
greatest at the edges).  LSP quantization produces a more localized, 
predictable effect with less work.

Monty

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