How much of what you hear is music, and how much is context?