Comparison of pronunciation scores in spoken language learning system

Abstract

An interactive language learning system should provide prolonged practice and feedback on individual problems. One of the challenges for such a system is to evaluate the learners' input speech for not only correctness but also appropriateness. In this paper we report on the state-of-the-art speech recognition technology for the assessment of pronunciation quality. Using nonnative data from Mandarin Chinese college students, we discuss five kinds of confidence based pronunciation scores such as likelihood, likelihood ratio, duration probability, posterior probability and modified likelihood ratio. Results obtained from the experiments show that these scores are helpful for students to improve their spoken language skills. © Springer-Verlag 2004.