Optimizing Teaching Strategies for American Vocal Singing Techniques Using Deep Reinforcement Learning in Music Education

Abstract

Music aesthetic voice education is the main way to cultivate students' music aesthetic and singing ability, in order to improve students' learning ability, this paper firstly analyzes the concepts and methods of music aesthetic voice teaching, and discusses the shortcomings of traditional music aesthetic voice teaching. Secondly, the wavelet threshold filtering is utilized to process the audio data and eliminate the noise in the audio. Finally, the curriculum and teaching strategies of music aesthetic voice were reformed through deep reinforcement learning techniques, so as to improve students' abilities. The results show that the maximum time used in recognizing 100 audio clips is 93.25s, which can correct students' singing errors in time. The teaching efficiency of the proposed method was high, increasing the previous 60% to 91%, enabling the students to fully grasp the knowledge explained. The scores of overall expressiveness in singing, vocal skills and emotional expression were 95, 96 and 95 respectively, and the optimized teaching strategy made the teaching of music beautiful voice more effective and promoted the development of music beautiful voice teaching.