International Symposium on Chinese Spoken Language Processing (ISCSLP 2002)

Taipei, Taiwan
August 23-24, 2002

Large lexicon construction for TTS system

Ben-Feng Chen, Guo-Ping Hu, Ren-Hua Wang

University of Science & Technology of China, Hefei, China

Lexicon is an essential part of Chinese Information Processing. In particular, compared with the basic lexicon, a large and perfect lexicon can effectively reduce the complexity and improve the precision of text parsing in TTS System. However, this special lexicon is hard to be constructed by either handwork or computer. This paper presents an approach to construct a large lexicon combining computer assistance and handwork, including the lexicon-iteration method of generating a large lexicon, and the lexicon-words selection that helps to improve the system. Based on this approach, we have constructed a large lexicon containing vocabularies about 200,000. And the experiments show that this large lexicon improves the efficiency of our system by 22.9% and the precision of word segmentation result by 19.0%.

Full Paper

Bibliographic reference.  Chen, Ben-Feng / Hu, Guo-Ping / Wang, Ren-Hua (2002): "Large lexicon construction for TTS system", In ISCSLP 2002, paper 55.