Will robot-composed music one day rise to the top of the charts? Probably not anytime soon, but researchers at Georgia Tech’s Center for Music Technology have a budding composer, Shimon, working in their lab.
The four-armed, marimba-playing robot represents seven years of artificial intelligence and deep learning research for Ph.D. student Mason Bretan. Bretan’s advisor, Gil Weinberg, created Shimon.
Bretan has taught Shimon how to “listen” to music by feeding the robot nearly 5,000 songs from multiple genres, from classical to pop, and more than two million motifs, riffs, and licks of music. The first robotic compositions were monophone improvisations over chord progressions Bretan provided.
Shimon’s new musical compositions demonstrate that, through deep learning, the robot knows how to approach music the way a human would. Bretan supplies four measures of music; Shimon grows this seed into a composition reflecting his understanding of the structure and rules of music.
“When we play or listen to music, we don’t think about the next note and only that next note,” said Bretan. “An artist has a bigger idea of what he or she is trying to achieve within the next few measures or later in the piece. Shimon is now coming up with higher-level musical semantics. Rather than thinking note by note, it has a larger idea of what it wants to play as a whole.”
Shimon is not the only musical robot created in Weinberg’s lab. He and his students have designed a robotic third arm for drummers and an interactive robotic companion that plays music from a phone and dances to the beat, among other devices.