Re: Adding a lossless mode to MPC SV8

> >Current spec only allows bandwidth scalability (normal for subband =3D
=3D3D=3D20
> >coders), but not scalable quality (AFAIK).
> >
> Hmmm ..... i hope Frank could consider to add this to the bitstream=3D2=
0
> specs, its an important thing to have it in the specs at=3D20
> least, even if=3D20
> its done in a later stage ....

The problem is possibly that this will always have a negative effect on =3D
bit
usage. You will always need to define in the bitstream, where exactly =3D
the
data for the improvement layer is.

> >but I'm not sure. Of course using a completely different=3D20
> coding method =3D3D
> >for
> >the lossless part, or error part, is also possible.
> >Menno
> > =3D20
> >
> Menno, yes, i guess the easier thing to do that was to intoduce a new=3D=
20
> 'mode' for that, leaving subband coding completely. BTW, the=3D20
> same way we=3D20
> could also add Vorbis encoding for the low bitrates ... LOL=3D20

I think this will leave quite complex synchronisation problems. My best
guess would be to losslessly encode the error signal between the lossy =3D
and
the original file. If you want to use different codecs for different
bitrates in the same file, you will always need to stack them in some
layered structure (eg. A low bitrate layer Vorbis, with an layer of MPC =3D
to
add up to the decoded Vorbis signal when a higher bitrate is required =3D
(and
then maybe a layer to make the file lossless)). The Vorbis filterbank
structure does not allow to switch on a frame to frame basis between =3D
codecs.
I'm not sure if this is possible with subband filterbanks, my guess is =3D
that
it is not, unless they have exactly the same delay, eg use the same
filterbank (duh).=3D20

Menno


http://mpc.corecodec.org