Study of Lecture Security for Hearing-impaired Students Using Speech Recognition Technology

开发了一种利用语音识别技术将教师的声音实时转换为字幕并显示给听障学生的字幕授课系统。在大学课堂上的实地测试表明，该系统可以将教师的语音转换为字幕，准确率约为93%。听障学生表示，讲课字幕系统使课堂比他们以前参加的课程更容易理解。不同于传统的使用语音识别重发方法的授课安全系统，不需要经过专门训练的人员重复语音，因此该系统具有较高的实用性。

A captioned lecture system that converts teachers' voices into captions in real-time by means of speech recognition technology and displays them to hearing-impaired students has been developed. A field test in university classes revealed that this system could convert the voice of teachers to captions with a high accuracy level of about 93%. The hearing-impaired students have stated that the lecture captioning system made the classes easier to understand than the classes that they had attended previously. Unlike the conventional lecture security system using speech recognition respeak method, there is no need for specially trained person to repeat the speech, and so the system has higher practicability.