Artificial intelligence is making its way to all possible domains. And we just cannot predict what will be the next innovation in this particular area. With the recent one, it has moved to the music field. A research team has designed a robot that has the potential to play and write its own music composition with the help of deep learning and artificial intelligence.
The robot, called as Shimon, has 8 sticks and 4 arms. It can play chords and harmonies on marimba. It also thinks in a similar way as a human musician, concentrating more on overall composition and less on the next note. The research team from the Georgia Institute of Technology, the United States, has provided input to the robot with almost 5,000 complete songs—from Miles Davis to Lady Gaga to the Beatles to the Beethoven—and more than 2 Million riffs, licks, and motifs of music.
Apart from providing a seed or the initial 4 measures that can be used as starting point to the machine, no individuals were involved in either the performance or the composition of the music. Mason Bretan from Georgia Institute of Technology said, “Once Shimon picks up the 4 measures provided by us, it makes its own series of concepts and creates its own piece. The compositions of Shimon denote how music looks and sounds when a robot utilizes deep neural networks to understand all it knows about music from the numerous human-created segments.”
The robot will be creating something distinct every time—the music which cannot be predicted by the researchers—as long as the team gives it a distinct seed input. Bretan, in the initial piece, fed the robot with a melody that included 8 notes. It achieved a sixteenth-note melody the subsequent time, which inclined it to create faster note sequences. According to the team, this dive in the musical quality of Shimon is because it uses deep learning that enables it to make a more coherent and structured composition.
The team also told that the debut of Shimon as a solo composer was presented in the Consumer Electronic Show keynote in a 30-minute video clip. It will also have its first live presentation at the Aspen Ideas Festival by June end.