The Application of Composition Technology Theory in College Music Teaching Based on Edge Computing under Digital Platform

Abstract

This essay examines how music theory is used in college and university music education to help students become more adept at analyzing and understanding a wide range of acoustic activities. In this research, we leverage the benefits of edge computing and digital platforms to create a compositional network structure for diatonic composition using the Markov model, two-way gated recurrent neural network, and curve fitting. This paper’s network architecture initially creates a Markov model to generate motivic melody for creating compositional knowledge rules. This model offers broad starting conditions for creating subsequent algorithmic compositions. Next, the style of the automatically gathered MIDI composition dataset is learned using a two-way gated recurrent neural network that can extract contextual note sequence information in order to create a prediction model. The test set achieves an accuracy of 88% on the proposed model by comparison experiments. The prediction model combines the input motivic melody to generate a one-part composition melody. At the same time, on the basis of the one-part composition melody, the relationship between the melodies of the two composing voices is studied, and a curve fitting method is used to model the two-part melody.