Here is a 28-second sample from Tool's track (-) Ions. I chose this to test first, as it gets an average bitrate of 384 kbps using Vorbis GT3b1 at q5.

Original sample is in FLAC format.

I encoded mp3 versions using only the commandline --alt-preset 128 in RazorLame. I used LAME 3.90.3 modified and 3.96b1.

FLAC and mp3 versions have ReplayGain information applied.

EDIT: This is my first attempt at testing LAME 3.96b1 against 3.90.3 modified. I noticed a different main artifact, depending on which version of LAME I used.

If anyone would like, they can post their comments about 3.90.3 and 3.96 encodings of this sampe in the LAME 3.96b regression examples thread.

So, what are we suppose to be looking for here? If you want to compare bitrates to Vorbis, then you should use VBR, not ABR.

Sorry, this is in reference to this thread.

The only reason I mentioned Vorbis at all is that is how I realized that this track is such a problem case. I should have made that clear in my first post. Sorry again.