公开(公告)号：EP1034533A1

公开(公告)日：2000-09-13

申请号：EP98952239.6

申请日：1998-10-09

申请人： Sony Electronics Inc.

发明人： CHEN, Ruxin ,  TANAKA, Miyuki ,  WU, Duanpei ,  OLORENSHAW, Lex, S.

IPC分类号： G10L5/04

CPC分类号： G10L15/142 ,  G10L15/148

摘要： A method and apparatus for a parameter sharing speech recognition system are provided. A model device (410) coupled to receive the output of a signal segmenter hosting a shared hidden Markov model produced by generating a number of phoneme models (600-603), son of which are shared. The phoneme models (600-603) are generated by retaining as a separate phoneme model any triphone model having a number of trained frames available that exceeds a pre-specified threshold. The generated phoneme models are trained, and shared phoneme models states (604-609) are generated that are shared among the phoneme models (600-603). Shared probability distribution functions (610-616) are generated that are shared among the phoneme models (600-609). Shared probability sub-distribution function (617-627) are generated that are shared among the phoneme model probability distribution functions (610-616). The shared phoneme model hierarchy is reevaluated for further sharing in response to the shared probability sub-distribution functions.