An Auxiliary Teaching System for Spoken English Based on Speech Recognition Technology

Abstract

Because the English language has always been inaccurate and seemed difficult to correct errors, this development has created a reputation based on improvements to the DWJ algorithm and HMM speech scores and correction mistake. In this paper, different signal characteristics are used using the DWJ algorithm: the Mel frequency cepstrum coefficient compares the standard speech library and the distance between the speech sample and the sample message received. The conversation deciphered the Viterbi code according to the HMM model, which was recognized and evaluated by posteriori probabilities. Finally, the professional data were used to fix the wrong phone to determine, score, and make repair mistakes. The results of the experiments show that the tests used in this article are reliable. The results of the experiment show that the standard English language proficiency in this article is reliable, which can provide students with timely, accurate, and objective assessment and teaching feedback, improving English language proficiency.