An Automatic Melody Generator based on the Constraint of Chord and Music Knowledge

Music is not an only accompaniment for narratives. It is also a representation medium of narratives. The theme of this paper is proposing a methodology of automatic generation of music as representation medium. First, we describe a system which automatically generates melody lines from a backing based on the constraint of chord and musical knowledge. We regard this as a system based on the harmony as musical structure in micro level. Next, as an extension of this idea, we propose a methodology of music generator based on the transformation of music structure. We define the more macro structure in music based on the idea of Genette’s narrative discourse theory and its expansion by computational modeling (expanded literary theory), and show a framework of music generation by the transformation of music structure.