Hi Gaby!
> > For my ears, Takehiro's scalefac_scale feature will not give better
> > results.
> > The quality of many tracks is a little bit lower. I have found no
> > tracks
> > with better quality. The file size is reduced, but for my opinion, the
> > quality is more important. So i think, setting this feature as default
> > is not a good idea. I'm using an electrostatic headphone, Stax SR-40.
> > It is a little bit old, but its quality is very high.
> >
> > Jack
>
> So it also means that with this feature and the same filesize, the
> quality could be improved each time.
I'm not so sure that the quality could be improved each time, even
if the resulting file is of the same size with or without it.
> This is why I think that when scalefac_scale is
> used in vbr, it would be good to change a little the vbr scale, or
> coefficients affecting the file size.
I have done something like that already for the old VBR code.
The quality will be adjusted by at least 1.25 dB.
Take a look in quantice.c, VBR_iteration_loop():
if (gfc->noise_shaping==2) {
adjust = Max(1.25,adjust);
}
> Is anyone ok with this or am I saying (telling? I never know wich one to
> use) something wrong?
In my opinion the scalefac_scale feature can improve the sound quality
with CBR encodings, when freed bits by this feature can be used by some
other bands. With VBR I fear we introduce only more noise, but you can
get the same effect by using a lower quality setting too. One point we
have to remember is, that VBR relies heavily on our psychoacoustic
model and when it fails it sounds bad. By this scalefac_scale feature
it seems we have to pay a higher price when GPSYCHO fails.
> Regards,
>
> --
>
> Gabriel Bouvigne - France
> [EMAIL PROTECTED]
> icq: 12138873
>
> MP3' Tech: www.mp3-tech.org
Ciao Robert
--
Sent through GMX FreeMail - http://www.gmx.net
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
