uh-huh! yup. Heard this before. Theoretically possible - but...let me just say I will believe it when I see it.

Programmers are still having a hard time getting single voiced sound to accurately convert to music - how are they going to get recognition of multi-voiced well enough to identify artist?