
How Do Computers "Hear" Music?
Michael Scott Asato Cuthbert opens MIT's Computational Music Theory and Analysis course by asking what it would mean for a computer to hear music, as opposed to merely process an audio signal. He draws out the gap between listening and hearing, framing music cognition as a problem of representation: what features of sound a machine needs to extract before any analysis of rhythm, pitch, or structure is possible. The session includes a live coding demonstration, likely using the music21 toolkit Cuthbert built at MIT, showing concretely how code can parse and manipulate musical data. It sets up the course's central question of how symbolic and computational methods can model musical understanding, laying groundwork for later sessions on analysis techniques. Aimed at students with interest in both music theory and programming, the class mixes conceptual discussion with hands-on demonstration rather than pure lecture.