摘要 |
PROBLEM TO BE SOLVED: To provide a reliability decision method for a voice recognition result of a voice recognition device. SOLUTION: The voice recognition device is equipped with a voice input means, a control unit, and a storage device, and converts a voice signal into a feature parameter series by a sound analyzing means and uses a sound model and a forward language model from the beginning of feature parameters by a forward search means to obtain a forward recognition result. Further, a sound model and a backward language model are used from the ending of the feature parameters by a backward search means to obtain a backward recognition result. According to the result of comparison between the forward recognition result and backward recognition result, a reliability decision means decides whether the section is reliable or not. COPYRIGHT: (C)2005,JPO&NCIPI
|