In some cases LAME 3.80b is producing worse quality files than LAME 3.70
Hi all,
Recently I received a file from Dromeas. He sent me the original wave file and
an MP3 encoded with LAME 3.80b VBR. The encoded file had very choppy sounds.
Must be some major VBR bug and I hope it is fixed by now. Viewing the graph, it
seemed to me that the MP3 file has been encoded in a way that the old FhG
encoder did...
Then I took a file with which the old FhG encoder produced very muffled,
underwater type sound. It was obvious enough to be detected with any speaker
system. I decided to test LAME 3.70 and LAME 3.80beta with that file. I
performed an ABC blind reference test with it. LAME 3.80b got much lower score
than 3.70. On an average, LAME 3.70 got 3.8 out of 5 and LAME 3.80b got 3.1.
Old FhG will get 2.5 highest. The difference is noticeable.
I hope you don't end up making an encoder like the old FhG. L3Enc may have a
lot of sophisticated routines, but some of them have drawbacks too. It's for
that annoying behavior with the files with "white sound" I gave up using that
encoder. Record something from the cassette, encode it with L3Enc and see what
happens! If anyone would like the file, please let me know where can I upload
it.
I hope to l work on the file that he has sent me and post the results later,
too.
- Zia
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
