> From: Dmitry Boldyrev <[EMAIL PROTECTED]>
>
> btw, here's my version of fastmdct
>
> #include "util.h"
> #include "l3side.h"
> #include "fastmdct.h"
>
...
Sorry - fastmdct.c and fiterbank.c are now gone...
This is one problem with not having a working cvs repository yet:
I just finished putting in Takehiro's new MDCT/filterbank code
which is noticably faster and has a delay of only 48 samples instead
of 528.
This is interesting because if Takehiro put his routines in a decoder,
we could get a total encode/decode dealy of a mere 96 samples. This
is kind of amazing since the MPEG papers all cite a minimum possible
delay of 1024-1056 samples. (but maybe not very important?)
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
