Music Performance Improvement Support System Using a Semi-Automated Instrument-Playing Robot with Real-Time Acoustic Analysis and Habit Visualization

This paper proposes an acoustic analysis system to help improve saxophone performance skills. The system combines direct support for performance movements by a robot with indirect support by presenting performance information. By sensing the performance audio and performing real-time acoustic analysis, the system presents the learner with information about their performance and their playing habits. The performance information presented to the learner includes pitch, volume, and playing timing. For performance habit analysis, a Markov model with pitch as the state and an internal probability parameter that indicates the quality of the performance evaluation as the pitch transitions are defined. In the experiment, we conducted a pilot study targeting experienced saxophone players and a beginner saxophone player to verify the effectiveness of the proposed system. The experiment showed that the MAE of the played pitch was significantly reduced by using the proposed system.