SPEECH RECOGNITION DEVICE, METHOD, AND SYSTEM

摘要

PROBLEM TO BE SOLVED: To recognize utterance contents of a speaker fast as a speech recognition. SOLUTION: A control unit 107 performs the speech recognition by matching the utterance contents input through a microphone 101 against standby words which are stored in data format of a hierarchial structure and belong to a higher layer and narrows down standby words belonging to the bottom layer derived from the higher layer. Based upon the narrowing-down result, it is judged whether or not the standby words belonging to the bottom layer need to be further narrowed down and when so, speech recognition is carried out by matching the utterance contents against standby words belonging to an intermediate layer; and the standby words belonging to the bottom layer derived from the layer are narrowed down to carry out speech recognition. COPYRIGHT: (C)2006,JPO&NCIPI