[Vorbis-dev] wavelet transform

[Dominik Zalewski]

Hello,

I've been thinking about decomposition in STN (Sine, Transient, Noise)
model. MDCT would stay for Sine then, DWT would be for Transients.

On 11/29/05, Monty <xiphmont at xiph.org> wrote:
>
>
>
> On Tue, Nov 29, 2005 at 10:54:46AM +0100, Dominik Zalewski wrote:
> > Hello,
> >
> > I am thinking about plugging in Discrete Wavelet Transform as
> > described in Vorbis I spec:
> >
> > 1.1.2. Classification
> > Vorbis I is a forward-adaptive monolithic transform CODEC based on the
> > Modified Discrete Cosine Transform. The codec is structured to allow
> > addition of a hybrid wavelet filterbank in Vorbis II to offer better
> > transient response and reproduction using a transform better suited to
> > localized time events.
> >
> > Could anybody tell me if it's good or bad idea?
>
> As a direct replacement for the MDCT?  Very likely a bad idea. For
> one, encoder doesn't know how to analyse the output of a DWT, and even
> if it did, it will probably give mych poorer energy compaction than
> the MDCT.
>
> Wavelets have their use, but not really as a substitute for the MDCT.
>
> Monty
>


--
Dominik Zalewski

Adam Mickiewicz University, Department of Mathematics and Computer Science
:: ul.Umultowska 87, room: B4-5
:: 61-614 Poznań
:: email: dominikz at amu.edu.pl, dominik.zalewski at gmail.com
:: mobile:  +48 692484801, phone: +48 618295333
:: gg: 2662959
:: www: http://www.staff.amu.edu.pl/~dominikz,
http://tri.wmid.amu.edu.pl, http://nlb.amu.edu.pl