Fourth European Conference on Speech Communication and Technology

Madrid, Spain
September 18-21, 1995

Methods Towards the Very Large Vocabulary Chinese Speech Recognition

Zuoying Wang, Jun Wu, Xi Xiao, Jin Quo

Department of Electronic Engineering, Tsinghua University, Beijing, China

Chinese is not an alphabetic language, and there is no convenient input method. Inputting by speech is one of the best ways to solve this problem. In this paper, we introduced our Chinese dictation machine, the JANET system, which is a very large vocabulary isolated speech recognition system. It could recognize all 1254 Chinese syllables as well as convert syllable strings into texts. The syllable recognition rate is 89% and the word accuracy is more than 95%. The system is implemented on PC and is commercially available now.

Full Paper

Bibliographic reference.  Wang, Zuoying / Wu, Jun / Xiao, Xi / Quo, Jin (1995): "Methods towards the very large vocabulary Chinese speech recognition", In EUROSPEECH-1995, 215-218.