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

Abstract

It is a challenge for the current music teaching system to carry out teaching according to the difference of score difficulty and realize automatic grading. Therefore, identifying the difficulty of music score according to pitch is the key to individualize music teaching resources. This paper summarizes and analyzes the problem of pitch feature extraction in music teaching. In the pitch extraction, the audio signal is divided into frames, and the feature matching of high-pitched content in music teaching resources is realized by smoothing the pitch sequence. In addition, the pitch feature extraction algorithm in MIDI music score files is proposed, and the pitch feature matching model is constructed. Finally, a music tutoring system based on pitch feature matching is designed, including a music score learning tool, overall structure of system, and interaction between teachers and students. Tutoring strategies include three main functions: learning suggestions of knowledge points, skills in practice and training, and learning path adjustment. This study is helpful to further improve the music teaching model and realize intelligent and personalized music learning.