On Sun, 5 Dec 1999, Ross Levis wrote:
> Unfortunately I'm running Windows rather than Linux. I will take that plunge once
>it is more configuration user-friendly. The -V4 switch in v3.57 (without -v which
>doesn't appear to be required). I noticed that prior Lame versions required -V5 to
>average close to 128kb/s whereas -V4 is now closer.
Yes, this is software in development. VBR is still a quickly changing
topic.
> I notice also that silence appears to be encoding at 64kb/s rather than 32 as
>specified, whereas Xing VBR (normal) encodes it at 48kb/s.
If the encoders are encoding this 'silence' at anything other then 32kb/s
then either it isn't silence, or both encoders are broken. :)
My expirence with Xing's VBR is that it often picks too low a bitrate. I
would expect that it uses a watered down psycoacoustic model to select
bitrate, insted of the slow lame approach.
> The major factor which concerned me was a frequency analysis I performed on a song
>encoded with -V5 (v3.57) which averaged around 112kb/s. It did not show anything
>over 16khz. I realise that without the -k option that bitrates below 128 are cut at
>16khz but surely with VBR there should be higher bit-rates and so higher freqencies.
If the cutoff is turned off and on within the file it produces an audible
effect. Therefor, the cutoff is selectd based on the -Vn setting and stays
throught the entire file. I think that -V5 and larger have the cutoff.
You can always disable the cutoff completely with -k.
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
