Application of Deep Learning-Based Speech System in Online Music Learning System

Abstract

Abstract Due to the continuous promotion of the application of the speech recognition technology of the twin network algorithm, the system performance has been continuously expanded, gradually becoming stronger, and voice control can be realized. Supporting target tracking in the twin network is a statistical-based recognition algorithm that solves the problems caused by insufficient sample size, inseparable linearity, hierarchical fracture, lack of a global view, etc. It has a wide range of applications, accurate classification, and is mostly used for speech recognition system. The research purpose of this paper is to use the sound frequency expansion measurement from the input data dimension of the audio and data system, and the contribution of the principal component analysis (pca) of the principal component value characteristic to the parameter recognition accuracy of the sound and data system can be maintained. At the same time, it greatly shortens the training time of the target tracking support twin network. This article focuses on the demand analysis of music course learning system software, designing software architecture and data structure, and briefly explains the implementation and testing process of the system. Directly use the mobile terminal WeChat applet architecture and server: Use JavaEE architecture and other technologies, including online learning, path training, and course management. Based on this, the research focus of this article is to design and implement a WeChat-based small program architecture, and test the solution algorithm.