【摘要】 本文介绍了自主研发的英语口语跟读题型机器阅卷系统,该系统采用以非特定人标准英语发音音素模型为基础设计的口语发音评测技术。文章分析了在大规模英语口语考试跟读题型中,机器阅卷取代人工阅卷的可行性和必要性。文中针对大学英语四、六级计算机网络考试试点测试的4010句考生录音数据,进行人工阅卷和机器阅卷实验对比分析,实验数据表明,两者的相关度为82.7%。本文还针对如何进一步提高机器阅卷的性能提出建议。更多还原

【Abstract】 A machine scoring system of repetition questions in English oral test is introduced in this article.The spoken pronunciation evaluation technology is based on speaker independent standard English phoneme pronunciation model.Feasibility and necessity of machine scoring replacing human scoring in large-scale English oral test is described in this paper and 4010 sentences of CET network English examination pilot are submitted for comparative analysis between machine scoring and human scoring.The experiment result shows that the correlation is 82.7%.This article also tries to make recommendations to enhance the performance of machine scoring.更多还原