Development of a system to recommend piano practice music utilizing difficulty level estimation algorithm

This study aims to develop a system to recommend an adequate difficulty level of music which suits a preference of a self-learning piano student. First, a user input a group of favorite music into a system and the system arranges the music according to the difficulty level. Next, the system analyzes the proficiency level of the user and recommends the music with the difficulty which is most similar to this level. With this system, users can select their practice music from the group of music arranged according to the difficulty level. In this paper, the authors used following procedure: first, a group of music from a piano manual which arranged several pieces of music according to the difficulty level are registered into a system; next, the system selects a piece of music based on the preference and the proficiency level of the user. The proficiency level is judged by whether the user can play the selected music or not. We proposed two algorithms to estimate the similarity of the difficulty level of music: One employs nearest neighbor algorithm and the other utilizes neural network.