Design of a Virtual Reality-Based Learning System for Spoken English

Abstract

Virtual reality is an important technology that is fast gaining global attention in different spheres of life particularly in the education sector. In view of this, this study designs a distance learning system for spoken English based on virtual reality, firstly, the overall design of the teaching system and the hardware and software of the system are designed, then a double-supervised signal convolutional neural network algorithm is proposed for the speech data recognition function of the system, and finally the testing of the system performance and the simulation analysis of the algorithm are carried out. The results show that the step response curve of the system designed in this study is gradually stabilized after 11s of operation, although there are certain fluctuations in the initial stage; the speaking scoring function of the system is more influenced by the sampling period T. When T is at 3 and 4, the speaking scoring speed of the teaching system is 33s~42s, which is significantly better than other intervals. The number of information submission and feedback was approximately the same and the interaction activity was very high after students used the system designed in this study, reflecting that student were more motivated to learn spoken English after using the system. The final loss rate using Goog Le Net is smaller and more convergent compared to the loss rate of the other three CNN models trained. The convolutional neural network algorithm constructed in this study has a very high accuracy rate in the recognition of English speech data, which is significantly better than other recognition models. To a certain extent, this study can provide guidance for the construction of English-speaking distance learning system, and more needs of users can be considered in future research.