>
> In the last episode (Aug 22), Jaroslav Lukesh said:
> > It should be maybe possible in wavelet transform, but not in discrete
> > cosinus. You should wait for wavelet encoder and decoder...
> >
> > Or you should use ;-))
> >
> > lame --decode x.mp3 - | mp3enc31 -sti -of x.small.mp3 -esr 22050 -qual 9 -bw 11025
>-br 64000
>
> I used to do that, but have found that lame doesn't do too badly on low
> bitrates nowadays:
>
> lame -a --abr 22 --lowpass 5.7 --resample 22.050 -S --mp3input input.mp3 output.mp3
>
> sounds very good. The most important value for me is the lowpass
> frequency; set it so the "twinkling" artifacts aren't too annyoing.
>
> Dan Nelson
> [EMAIL PROTECTED]
I tend to agree: With a proper lowpass filter setting, LAME can get
pretty close to the quality of mp3enc at low bitrates.
One reason mp3enc might still have an edge is that for low bitrate
stereo encodings, it uses intensity stereo, something LAME
still does not have. Also, just a few days ago Robert discovered
a bug which was causing stereo MPEG2 encodings to not use
mid/side stereo as often as possible.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
