Hi all,
i'm back from holiday and jumped into a bit of coding.
I'll write a little report on my MDCT findings in the near future - but
for now i've found out that the MDCT that I hand unrolled (the thing currently
in LAME) is pretty good.
I implemented an MDCT based upon an size 9 (ie N/4) complex FFT (similar
to bosse's mdct for his project). This is supposed to be a very efficient
implementation. After optimizing and fiddling and trying lots of different
things I found that it was pretty much the same as what LAME already has.
I'll post my findings and code later just in case anyone else wants to
have a shot at it.
later
mike
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
