Chinese oral English quality assessment method independent from text

Abstract

The invention relates to a Chinese oral English quality assessment method independent from text. A standard English pronunciation acoustic model of the Wall Street Journal and Chinese oral English pronunciation data whose expression is correct but pronunciation is not standard are input to establish an adaptive pronunciation recognition model of oral English via training; the standard English pronunciation acoustic model of the Wall Street Journal and Chinese oral English pronunciation data whose expression is correct and pronunciation is standard are input to establish a standard adaptive assessment model of oral English; input oral speech data of a person to be assessed is identified by utilizing the adaptive pronunciation recognition model of oral English; and the identified oral speech data is processed by the standard adaptive assessment model of oral English, and the quality and score of the oral speech of the person to be assessed is output. The method based on acoustic mixed English models and independent from text can solve the problems in assessing the oral English quality of Chinese by a computer system.

Description

technical field The present invention relates to speech recognition technology, speech signal processing technology and natural language processing technology, it is a kind of method that uses computer system to the Chinese spoken English quality that the text is irrelevant to evaluate. Background technique Spoken English quality assessment technology can not only help English learners understand their own oral English pronunciation level, but also help to find problems in oral English learning, so as to improve their own oral English pronunciation and improve their own oral English learning efficiency. Moreover, the oral English quality evaluation technology can also be applied to the oral English test to reduce the workload of manual evaluation in the traditional oral English test and ensure the objectivity and accuracy of the oral English test evaluation. Existing oral English quality assessment techniques are mainly divided into text-related oral English quality assess...

AI Technical Summary

Problems solved by technology

The present invention is a text-independent Chinese spoken English quality assessment method based on the English acoustic mixed model, which solves the problem of using a computer system to evaluate the Chinese spoken English quality. The English acoustic mixed model of the method includes an English spoken English adaptive voice Recognition model and a self-adaptive evaluation model of standard spoken English. When evaluating the quality of Chinese spoken English, first use the self-adaptive speech recognition model of spoken English to perform speech recognition on Chinese spoken English to improve Chinese spoken English speech recognition rate; and then use the English standard oral English adaptive evaluation model to evaluate the quality of Chinese spoken English obtained from speech recognition to ensure that the evaluation quality has good credibility

Images