Analysis and utilization of MLLR speaker adaptation technique for learners' pronunciation evaluation
Analysis and utilization of MLLR speaker adaptation technique for learners' pronunciation evaluation
复制标题
DOI：
10.21437/interspeech.2009-218
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
Dean Luo;Y. Qiao;N. Minematsu;Yutaka Yamauchi;K. Hirose
中科院分区：
文献类型：
--
作者：
Dean Luo;Y. Qiao;N. Minematsu;Yutaka Yamauchi;K. Hirose
在本文中，我们研究MLLR说话人自适应的效果和问题时，应用于发音评估。在两个公开的日本学习者英语发音数据库上进行了自动评分和错误检测实验。正如我们所预料的那样，过度适应会导致发音准确性的误判。在这些实验之后，提出了两种新的方法，强制对齐GOP评分和正则化MLLR自适应，以解决MLLR自适应的不利影响。实验结果表明，该方法能更好地利用MLLR自适应，避免过适应。索引术语：计算机辅助语言学习（CALL），说话人自适应，发音评估，发音优度（GOP），最大似然线性回归（MLLR）
In this paper, we investigate the effects and problems of MLLR speaker adaptation when applied to pronunciation evaluation. Automatic scoring and error detection experiments are conducted on two publicly available databases of Japanese learners’ English pronunciation. As we expected, overadaptation causes misjudge of pronunciation accuracy. Following these experiments, two novel methods, Forced-aligned GOP scoring and Regularized-MLLR adaptation, are proposed to solve the adverse effects of MLLR adaption. Experimental results show that the proposed methods can better utilize MLLR adaptation and avoid over-adaptation. Index Terms: Computer Assisted Language Learning (CALL), speaker adaption, pronunciation evaluation, goodness of pronunciation (GOP), maximum likelihood linear regression (MLLR)