Estimation of Musical Impression by Fuzzification Musical Impression Value in Consideration of Individual Difference

Abstract

In this paper, we propose an estimation method of musical impression value based on human's KANSEI as the elemental technology for the musical retrieval system. The main focus is to estimate the fuzzy impression value in order to consider the difference in musical impression among audience's individuals. We estimate the fuzzy impression value using linear programming with the fluctuation information which indicates time structure of musical piece. In experiments to estimate the fuzzy impression value, the results showed that this method is reflecting the difference by audience's individuals.