Modeling of music recommendation methods to promote the user's singing motivation - For next-generation japanese karaoke systems

Abstract

This study attempted to build a model that recommends music choices to encourage karaoke-system users to sing by using data about the music preferences and inner characteristics of each user. First, we conducted an auditory experiment in two phases. Additionally, we analysed the acoustics and lyrics of music pieces. Using these data, we built a map of the music based on user impressions, and used this map to reveal the relationship between the user's most favourite music piece and the music piece that a user was highly motivated to sing. Thus, we were able to establish a basic model of the system that recommends the music piece a user would be highly motivated to sing. © 2013 Springer-Verlag Berlin Heidelberg.