摘要 |
PROBLEM TO BE SOLVED: To provide a speech recognizer that recognizes a specified word from speech uttered by a user and controls an apparatus, having a recognition dictionary which hardly causes misrecognition. SOLUTION: The speech recognizer 1 recognizes speech by verifying feature values extracted from speech signals of the speech uttered by the user with the feature values of words preregistered in the recognition dictionary. This speech recognizer has a general dictionary with large vocabulary, and also has a similarity calculation means for finding similarities with the words included in the general dictionary with respect to the words registered in the recognition dictionary by using a predetermined algorithm, and a similarity presentation means for presenting the fact that the similarities are high to the user concerning the words registered in the recognition dictionary having great similarities with the words included in the general dictionary. COPYRIGHT: (C)2005,JPO&NCIPI
|