> Date: Sat, 1 Apr 2000 17:40:39 +0200 (CEST)
> From: Tomasz Motylewski <[EMAIL PROTECTED]>
>
> I have seen that Lame does not support 11.025, 12 and 8 kHz sampling. I have
> lots of streams encoded at 11.025 (teleconferences), when I have tried
> resample option effects were unsatisfactory at low bitrates. I think it
> should be possible to compress it to about 8 kbps.
>
have you tried resampling with sox? it gives much better quality,
especially for upsampling.
> I have started by putting in the following patch, but I have realized that
> there are more modifications needed. Could you tell me what are the other
> things which should be changed?
>
> Best regards,
> --
> Tomasz Motylewski
>
Thanks for the patches - I just added them. With a few other tweaks
(syncword appears to be FFE instead of FFF?, and a couple of valid
samplerate checks) lame now seems to be producing mpeg2.5 streams!
(mpg123 reports many playback problems. But maybe this will be enough
to get you started.
In addition to the playback bugs, the psycho acoustic information
needs to be updated:
1. bark values of each band: (can be computed, see #define BARK)
2. partition band tables (i.e. which FFT coefficints are in which partition band)
3. scalefactor band tables (which partition bands are in which scalefactor bands)
As a first pass, the psycho acoustics for MPEG2.5 can just use the
data for MPEG2. (this is done now). All of the data for 1,2 & 3 will
eventually be replaced by formulas.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
