作者: L.R. Bahl , P.V. deSouza , P.S. Gopalakrishnan , D. Nahamoo , M.A. Picheny
DOI: 10.1109/ICASSP.1991.150308
关键词:
摘要: The authors present an automatic method for modeling phonological variation using decision trees. For each phone they construct a tree that specifies the acoustic realization of as function context in which it appears. Several-thousand sentences from natural language corpus spoken by several speakers are used to these Experimental results on 5000-word vocabulary speech recognition task presented. >