I wouldn’t be surprised if a *sound* registry using shazzam/echonest/pandora audio analysis isn’t proposed at some point.