Put a very hard filter around the vocal range of sound, hopefully that would be enough to distort the music so that the algorithms can’t identify it.