CguAlignBuilder: An audio book construction technology for speech-text synchronization with the application to a language learning website

• Main Authors: Po Ting Chen, 陳柏廷
• Other Authors: R. Y. Lyu
• Format: Others
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/36095084233598165990

碩士 === 長庚大學 === 資訊工程學系 === 101 === This research subject is CguAlign, which is speech-text synchronization. It is an audio book, auxiliary system, on website for people to use it. The fundamental principle is using speech-text alignment of speech recognition technology, web design technology, and construct and manage server. Using these technologies build the speech-text synchronization’s audio book on website. We call CguAlignBuilder. We use HTK, SailAlign, and CguAlign (three kinds of technologies) to assist the audio book, and these technologies also can improve timing limited problem. In information structure and file format, we use Transcriber, Lyrics, JavaScript, and etc…, and we also add in our own manufacture system for assisted. That includes Apache, FFMPEG, and managed text and voice file technology. And then, we count how long will take for transfer running this system. We test thirty one files for this research, which include seventeen English files and fourteen Chinese files. The file will take eighteen hours and thirty five minute for transfer. On IntelCorei5 and DDR3 4G computer, the file will take one hour and forty minute for transfer. This system already set up on website for people to use it.