Devin pointed me to an incredible video. Apparently, German computer scientists have figured out how to split a musical recording into its component notes, allowing you to manipulate a digital recording of a piece as easily as if it were just a MIDI recording track. This means that you could generate a cappella versions of your favorite song, or make an artist sing in harmony with herself, or simply fix a one-note recording error, all without having access to the original master tracks or doing any additional recording. You can see the (admittedly corny) video for a great overview of what they’ve accomplished and some great examples of what this technology makes possible.