基于多特征融合的汉英口语翻译自动评分方法研究  

Research on automatic scoring method of Chinese-English oral translation based on multi feature fusion

作　　者：王婧锦 Wang Jingjin(School of Foreign Languages,Shaanxi University of Traditional Chinese Medicine,Xianyang,Shaanxi 712046)

机构地区：[1]陕西中医药大学外语学院,陕西咸阳712046

出　　处：《现代科学仪器》2021年第2期258-261,共4页Modern Scientific Instruments

摘　　要：汉英口语翻译自动评分不仅仅可以提高评分的效率,同时也可以确保评分的公正客观性。本文提取了汉英口语翻译自动评分的相似度特征、句法特征和语音特征,得到了用于自动评分的11个特征。以某大学英语专业口语翻译课程考试所采集的语音信号为例,最终获得了包含8个特征的汉英口语翻译自动评分模型。结果表明,在去除特征之后,汉英口语翻译自动评分系统的性能得到了一定程度的改善。本论文的研究对汉英口语翻译自动评分研究具有一定的参考机制。Automatic scoring of Chinese English oral translation can not only improve the efficiency of scoring,but also ensure the fairness and objectivity of scoring.This paper extracts the similarity features,syntactic features and phonetic features of automatic scoring of Chinese English oral translation,and obtains 11 features for automatic scoring.Taking the speech signals collected from the oral translation course examination of a college English major as an example,an automatic scoring model of Chinese English oral translation with eight features is obtained.The results show that the performance of the automatic scoring system has been improved to a certain extent after the feature is removed.The research in this paper has a certain reference mechanism for the study of automatic scoring of Chinese English oral translation.

关 键 词：多特征融合 自动评分 相似度特征 句法特征 语音特征 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]