Multi-Regression Analysis of Music Impressions for Music Evaluation

In this paper, we propose a music evaluation method. In this method, impressions of given music are represented by features by value, so that the music impressions are processed on computers. To obtain the music impressions, we experiment music impressions by using a semantic differential method. Then, we construct several models based on the experimental results with multiregression analysis. By using these models, music features are uniquely digitalized. A multi-agent system is expected to use the music feature. Each agent of the multi-agent adopts a different model to cooperate between agents. Then, users can understand their music preference based on the music features values. 1. は じ め に