Hi Everyone,

I have been very surprised at the responce the LAME 3.0
(www.sulaco.org/mp3) posting had!  Sorry for the frequent updates,
(3.0, 3.01 and now 3.02 in two days) but a lot of people have tried it
and emailed comments, bugs and fixes.


Here's a summary:

Warren Toomey tracked down a subtle variable initization bug 
that only showed up under FreeBSD.  

Several people found that I had broken mono encoding.  fixed in 3.01

Jan Rafaj sent me a sample which showed a flaw in ms_stereo.  LAME
would sometimes switch to ms_stereo even if the two channels
were being encoded with different block types.  (fixed in 3.02)
This sample also shows the superiority of the FhG short_block switching
criterion - something that needs work in LAME.

Jani Frilander found a bug in the bit reservoir routines
that shows up if you have a lot of frames of all zeros.  (fixed in 3.01).  

Scott Manly send a patch to allow seeks on the input file,
(for front end programs that look for header information?)
added to 3.01.  

Mike Oliphant increased MAX_NAME_SIZE to allow artist/disc/track format.
(set to 300 in 3.01)

At 128kbs or less, I remove the MDCT coefficients in the last
scalefacor band, as does the FhG demo encoder v3.1.  
This amounts to a 16kHz cutoff.  If you dont like this you can
now disable it with -k.  (3.02)

Reply via email to