Those songs are identical in every pragmatically meaningful way, but they still won't be bitwise identical.

You'd have to do aural spectrum comparisons, with appropriate tolerances.

Which is why no one bothers.