Optimization Design of Intelligent Pronunciation Assessment System Based on Speech Recognition Technology in Higher Vocational English Classroom Teaching under the Background of Industry-Teaching Integration

Abstract

The in-depth development of the integration of industry and education requires that English teaching in higher vocational colleges and universities be equipped with more objective and convenient methods and means for students' speech recognition, detection and assessment. In this paper, a computer-aided English pronunciation assessment system is established by utilizing speech recognition technology and combining the characteristics of English pronunciation with the verification of speech segments, the cutting of language signals and the assessment of pronunciation. MFCC is selected as the voice feature parameter of the proposed system to meet the needs of pronunciation model training. For accent assessment, the neural network-based GOP method is introduced as a deep learning-based accent assessment method to construct a theoretical modeling method for the English pronunciation assessment system. Compared with the traditional methods, the modeling method proposed in this paper consistently maintains an average error between 0.001-0.1 on the automatic assessment of English spoken pronunciation quality, which demonstrates a superior performance of automatic assessment of spoken pronunciation quality.