The Study And Design Of Oral English Scoring System

In recent years, Computer assisted language learning is a discipline developed rapidly, and has attracted more and more attention from scholars. It mainly promotes simple language teaching activities with the support of computer and information technology. But in fact it rarely can be used in the teaching practice of oral English. The main reason is the lack of the evaluation of learners’ oral English pronunciation and the feedback of information correction. And effective and correct feedback is vital to improve users’oral English pronunciation as well as the evaluation of the system’s performance, while it backfires if the learners’correct pronunciation is regarded as error. With the preliminary studies of speech recognition technology, scoring method and the theory of information feedback, this paper puts forward an improved method based on the HMM posterior probability score.It means that it no longer uses the only standard reference model as the basis of scoring and judging true or false, but takes advantage of the average level of the standard pronunciation in the whole corpus. The method can not only reduce score limitation caused by different person’s standard pronunciation, but also reduce the misjudgment rate the system has on pronunciation error detection, so as to increase the effectiveness of correction information. And with the establishment of the database which is made up of experts’opinions, and depends on the type of common spoken English pronunciation errors, and combining the correct information, it can help learners to improve spoken English effectively.At last, this paper designs an automatic scoring system of spoken English, which makes those functions such as recognition of users’pronunciation, grading, error judging, correcting feedback, and other functions come into effect with the help of Sphinx4. Finally through experiment and test, it was confirmed that the system could effectively improve the system’s scoring performance and users’pronunciation ability after introducing the average pronunciation level.