Co-creative Composition Using Serial Chord Generation with Reinforced Learning

摘要

In this paper, we proposed the Co-creative Composition in which the musical note agent decides the next position on the musical sheet-like environment, interacting with other agents. The agent decides its action with reinforced learning according to Sensory Consonance Theory. We verified the generated cords, compared with music theory and using psychological experiment Results showed that the generated chords had similarity to the music theory, and showed that the subjects felt clarity of sound to them.