Using Pitch Feature Matching to Design a Music Tutoring System Based on Deep Learning

Area of Science:

• Music Education Technology
• Signal Processing
• Artificial Intelligence in Music

Background:

• Current music teaching systems struggle with individualized instruction based on score difficulty and automatic grading.
• Personalized music learning requires accurate identification of music score difficulty, with pitch being a key factor.

Purpose of the Study:

• To address the challenge of individualized music teaching by developing a pitch feature extraction algorithm.
• To construct a pitch feature matching model for music score analysis.
• To design an intelligent music tutoring system that personalizes the learning experience.

Main Methods:

• Audio signals are segmented into frames for pitch sequence smoothing and feature matching.
• A novel pitch feature extraction algorithm is proposed for MIDI music score files.
• A music tutoring system is designed, incorporating a learning tool and teacher-student interaction functionalities.

Main Results:

• The study successfully extracts pitch features from music scores to determine difficulty levels.
• A functional music tutoring system is developed, demonstrating the efficacy of the pitch feature matching model.
• The system provides personalized learning suggestions, practice guidance, and adaptive learning path adjustments.

Conclusions:

• The proposed pitch feature extraction and matching methods are effective for identifying music score difficulty.
• The developed music tutoring system facilitates intelligent and personalized music learning experiences.
• This research contributes to improving music teaching models and advancing adaptive learning technologies in music education.