Musical Expression Generation Reflecting User’s Impression by Kansei Space and Fuzzy Rules

This paper proposes a generation method of musical expression based on a performer’s impression. The musical expression generation method consists of two procedures: image estimation and derivation of parameter values ofmusical expressions. In the image estimation procedure, an adjective, i.e., an image word, is mapped into the Kansei space. In the parameter values derivation procedure, parameter values of musical expression, tempo, volume and length of a note, are obtained by mapping from the Kansei space to the parameters’ space by fuzzy inference. The validity of the proposed method and the influence of music genres on musical expression generation are confirmed by subject experiments. From the experimental results it is found that the proposedmethod successfully generates musical expression reflecting impressions and musical expression in several genres.