----- Original Message -----
From: Andree Buschmann <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Friday, June 02, 2000 5:26 PM
Subject: Re: [MP3 ENCODER] MPEG plus
> it's based on the strucure of mpeg1-layer2 and therefore is a
> subbandcoder. there were some enhancements in lossless coding (intense
> use of huffman-tables for quantized samples, more efficient coding of
> scalefactors), some enhancements in the psychoacoustic model (like
> 'clear voice detection' which prevents voiced speech from becoming
> distorted or nonlinear spreading function) and some things that were not
> supported by layer2 (mid/side-encoding, adaptive noise shaping in
> subbands). the codec is recommended to be used in vbr-mode (quite low
> average bitrate) and seems to offer more 'stable' quality in comparison
> to other codecs. as i've been told with special hard-to-encode signals
> (like 'fatboy slim - kalifornia') it performs better at vbr with avg of
> about 140 kbit/s than mp3-encoders with 192 kbit/s...
>
If I understood well, this was a student project, and thus you don't plan to
make money with it. So you could perhaps share your work with us. ( ie
explanations, source code, or contributions to Lame/Vorbis?)
I'd like to know how you achieve such a difference in final bitrate compared
to mp3. Lame also uses huffman tables and M/S stereo. So the only thing left
is the psychoacoustic model. Woud it be possible to have further
explanations?
Note: I didn't tested MP+ yet, but explanations would interest me more than
just using it.
Regards,
--
Gabriel Bouvigne - France
[EMAIL PROTECTED]
icq: 12138873
MP3' Tech: www.mp3-tech.org
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
