Fuzzy Evaluation Model for Oral English Test Based on Automatic Scoring System

Abstract

With the rapid development of the mobile Internet, the design of a fuzzy evaluation model for spoken English based on an automatic scoring system has become a research hotspot in the field of spoken English testing at home and abroad. This article focuses on the current spoken English evaluation problems in English learning and analyzes the characteristics of English pronunciation. On the basis of the characteristics of speech prosody, a speech evaluation model based on fuzzy comprehensive evaluation is proposed. At the same time, the method is transplanted and an automatic spoken evaluation system based on the Android platform is implemented. One of the key points of the automatic labeling system is the selection and extraction of prosodic features and acoustic parameters. This paper improves the automatic labeling of accents by improving the corresponding rules of prosody and acoustic parameters. The introduction of the Co-training algorithm greatly reduces the amount of manual labeling, and realizes automatic labeling of prosody with minimal labeling. In this paper, a detailed performance analysis and experimental comparison of the improved voice evaluation model are carried out through Matlab simulation experiments. The improved model is highly efficient and practical. Finally, the improved voice evaluation model was transplanted to the Android platform, and a voice evaluation system based on the cloud platform was designed and implemented. In this paper, the evaluation effect of the system was tested by experimenters. The experimental test showed that the agreement of the test expert evaluation reached 95%. Therefore, the model established by selecting characteristic parameters based on prosodic features has a better evaluation effect.