HOW: I used hidden microphones to pick up the voices, then fed the signal into a mixer to amplify it, then into the software program Max. Max performs a series of algorithms to clean and separate human voice from the background noise, then to isolate various pitches I’m anticipating. This data is then sent to Flash using a Java server called Flosc. Flash then uses the amplitude and pitch data to generates graphic "words" which were projected upon the wall.