发音属性优化建模及其在偏误检测的应用(Speech attributes optimization modeling and application in mispronunciation detection)

Minghao Guo (郭铭昊), Yanlu Xie (解焱陆)


Abstract
近年来,发音属性常常被用于计算机辅助发音训练系统(CAPT)中。本文针对使用发音属性的一些难点,提出了一种建模细颗粒度发音属性(FSA)的方法,并在跨语言属性识别、发音偏误检测中进行测试。最终,我们得到了最优平均识别准确率约为95%的属性检测器组;在两个二语测试集上的偏误检测,相比基线,基于FSA方法均获得了超过1%的性能提升。此外,我们还根据发音属性的跨语言特性设置了对照实验,并在上述任务中测试和分析。
Anthology ID:
2020.ccl-1.7
Volume:
Proceedings of the 19th Chinese National Conference on Computational Linguistics
Month:
October
Year:
2020
Address:
Haikou, China
Editors:
Maosong Sun (孙茂松), Sujian Li (李素建), Yue Zhang (张岳), Yang Liu (刘洋)
Venue:
CCL
SIG:
Publisher:
Chinese Information Processing Society of China
Note:
Pages:
66–76
Language:
Chinese
URL:
https://aclanthology.org/2020.ccl-1.7
DOI:
Bibkey:
Cite (ACL):
Minghao Guo and Yanlu Xie. 2020. 发音属性优化建模及其在偏误检测的应用(Speech attributes optimization modeling and application in mispronunciation detection). In Proceedings of the 19th Chinese National Conference on Computational Linguistics, pages 66–76, Haikou, China. Chinese Information Processing Society of China.
Cite (Informal):
发音属性优化建模及其在偏误检测的应用(Speech attributes optimization modeling and application in mispronunciation detection) (Guo & Xie, CCL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.ccl-1.7.pdf