I just checked into CVS a fix of the mid/side maskings.
The formula was using ratios instead of the true thresholds,
which I didn't realize was a mistake until looking at Menno's
AAC code.
I did not yet put in any of the other suggestions from Menno
The biggest change would be the L & R maskings:
It allows masking in one channel to mask noise in the other channel
at high frequencies. At low frequencies, there is a limit to how
much cross channel masking is allowed (controled by the MLD function).
It sounds like a good idea, but I want to find at least one
test case where it improves things before adding it. Usually
small changes to the masking has very little effect on the output.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
