Ahhh I see. Thanks for the link.
That would work well if the person being tested were able to have control over the player so they could switch back and forth quickly.
I would imagine that the differences are much more apparent with music that the listener is intimitely familiar with.
The only one I've used is winabx. You can set begin and end marks to compare a short passage, and certainly if you set the test up yourself you can use familiar music.
If you are training yourself to tell the difference you might want to start with some classic cases known to expose problems like "castinets"
Here are a bunch of samples considered "obvious," with descriptions of the artifact and the sample available in "bad" (perhaps Xing 128kb/s), "good"(lame insane), and lossless.
Artifact Training Page