>
> Pierre/Hurgon-
> I don't know about the first half, but I'm prettty sure I know about the second half-
> This is basically what you said...
>
> >c:\lame\exe\lame -h -b 96 test2.wav file1.mp3
> >Resampling: input=44kHz output=32kHz
> >Using polyphase lowpass filter, transition band: 10968 Hz - 11355 Hz
> >
> >c:\lame\exe\lame -h -b 96 --resample 44.1 test2.wav file2.mp3
> >Using polyphase lowpass filter, transition band: 12270 Hz - 12803 Hz
> >
> >
> >Why the transition bands aren't the same ?
>
> First time- You just told Lame to encode the file at 96kBit/sec. Lame
> thinks 96kbps is not enough data to make a good copy of a stereo
> 44.1kHz file. So it automatically resampled the WAV & was a little
> more "vicious" with the LowPass Filtering. At low bitrates, Lame has a
> tendency to add some LPFing.
>
> Second time- You told Lame to encode the file at 96kBit/sec, & you
> also told it to set the sample rate to 44.1kHz When you set the
> options in the second command line, you override what Lame was going
> to do - That's to resample the WAV before encoding. So it just
> continued to encode at 44.1kHz.
>
> >course the same size, but the second mp3 sounds better (sure, of
> >course) than the first one, without quality loss.
>
> Have a close listen to both encodings & if they don't sound very
> 'spatial', then that's why. Try both the command lines again, this
> time adding the '-k' switch... That's the 'full bandwidth' switch
> from memory, but look in the documentation to be sure. When using
> full bandwidth, Lame won't/shouldn't do any LPFing.
>
> Shawn
Here's why lame does what it does:
96kbs 44.1khz: compression of 14.7x
96kbs 32khz: compression of 10.7x
The 96kbs/32khz file be will have better quality encoding (since less
compression) but the encoding source is lower quality (32khz).
The 44.1khz will have not as good encoding quality, but
the encoding source is higher quality.
Compressing by a factor of 14.7 is pushing the limits of
mp3. It would sound pretty bad without the lowpass filter.
The amount of lowpass filtering is based on the compression
ratio. It's a simple formula that was tuned for the low bitrates,
so it may not be doing the right thing at 96kbs.
If the 96kbs/44.1khz really sounds better, I would think the
96kbs/32khz, with the same amount of filtering (--lowpass 12.5) would
sound the best.
which is optimal? good question :-)
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
