I'd like to try to tell when live audio input is a "song" -- for example Happy Birthday -- as opposed to speech or random noise. Can anyone suggest any further reading on strategies for doing this. Thanks.