Hope someone can comment on this:
I tried the Windows compilations of Lame since v. 3.32 and got the strange result that
every version from 3.33 on (tried up to 3.36) was 2.5 times slower than 3.32.
This on a Celeron CPU running at 450 MHz. Lame 3.32 runs about 2.4x real time and the
later versions all slightly under 1x, same parameters set, e.g. 128 k, VBR on or off
same result.
In the release notes, there is nothing mentioned that could affect speed.
How can that be explained ?
About quality:
I tried Lame mainly because Audioactive (they licensed the Fraunhofer codec), while
really good w. most music sources, was unable to encode one tune I cut from MTV (Oh
Carolina) in that it ran over it twice as fast and produced something sounding like
56kbps. I also noticed some warbling behind vocals from clear CD sources at 128k. None
of these problems occured w. Blade and Lame, same bitrate. I couldn't get to hear any
difference to CD w. lame 3.32 at 128kbps and VBR on, at least w. the music sources I
have. So I can't verify how the improvements of the later versions really are.
Regards
Rolf
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
