Speech recognition is getting there...but ever so slowly. Of course, if you're mean to your recognition engine, the results can be pretty bad. Don't talk with food in your mouse, don't speak too loudly, don't slur things, etc. But when you give it music...well, it can actually be pretty poetic in a screwed up random kind of way.

I fed the MS Speech Recognition system a Ben Folds Five song (Magic). After the first verse and refrain, this is what it came up with:

From my own land and and a two-man a plan Pan am up the zero there's a good man a tan and And they uses new as is today and you knew In her hand who so a and stuck And they a gang and gave a ten Z. A. A.

Speaking Japanese to came up with similar results, though with fewer rhymes, as to be expected. Mind you, this is on very little training and it only gets about 85% of my words even when I speak english normally.

See, it's not that I have too much free time, but that I'm just crazy...