>
> > My results so far are 2X realtime (twice as fast as realtime)
> > on a 600MHz 21164 system and 3X realtime on a 600MHz 21264.
> really?
> it's about 2.75X on a P166 (freebsd).
>
Mike, on your web page you say "Speed is 5 times faster than original
ISO code. About 115 seconds to encode a 40 second file on P166" Thus I
think you mean 2.75X *slower* than realtime, while Maurice Marks was
quoting 3X *faster* than real time?
I was reading FhG's comments on MPEG-AAC. Most of you probably
already knew this, but I'll summerize: Same quality as
layer-3 with 30% less bandwidth. Some new psycho-acoustics,
(but from what I read I think this mostly helps speech encoding,
not music?) Which would mean that most of the improvement
comes from the mathematics: better quantization, more efficient
bit-stream format and simpler filter bank (a single MDCT instead
of the unnecessary sequence of two fitler banks used in layer-3).
Anyway, the most interesting thing was in the FAQ:
Q: Will there be a free decoder?
A: No. There is a public decoder source code
available via MPEG but even when you use
this public code, royalties have to be paid to
the patent holders. For details on licensing,
consult [EMAIL PROTECTED] Additionally, MPEG
provides an encoder source code. This
encoder is mainly intended to facilitate
decoder development and does not provide
AAC-like quality.
No free players probably means it wont replace layer-3 any time soon...
Mark
