An Innovative Study on the Interaction Mode of Artificial Intelligence-Enabled Music Classroom

Abstract

AI music creation software can inspire students to create music, so that they can not only learn music theory and practical skills in the process of creating music, but also form a harmonious classroom interaction mode of teacher-student interaction, student-student interaction, and learning individual and teaching intermediary. In this regard, the study designed a music generation model based on Transformer-XL to generate higher quality music. The model consists of a fragment-level recursive mechanism and a new relative position encoding scheme, which is more conducive to improving the harmony of the generated samples. Comparative evaluation of this paper's model with some more advanced mainstream music generation models, such as MiDiNet and MusicVAE, in terms of objective metrics and subjective listening tests shows that the Transformer-XL model proposed in this paper outperforms other generative models at both the subjective and objective levels, and can generate higher quality music. The interactive effect of classroom based on AI music creation is evaluated based on the Flanders Interaction Analysis System, and the evaluation results reflect that AI music creation increases the interest and participation in the classroom, allows teachers and students to have a more diversified and interactive way of learning, and injects new vitality into the traditional music teaching, which makes the music classroom become more vivid and efficient.