Speech recognition system

摘要: PROBLEM TO BE SOLVED: To actualize an excellent speech recognition interface by limiting words and documents as objects of performing the recognition. SOLUTION: A analysis part 106 performs a noise process takes inputted through microphone 106. 107 collates input with analytic result calculated using dictionary 105 sound model 108 in sequence to calculate closest word 105. In this case, switching 103 selects one dictionaries or switches them for according contents information 102. For example, plural are stored on memory card ROM 104, when is performed, only necessary transferred RAM perform recognizing process. Namely, be recognized limited performed. COPYRIGHT: (C)1998,JPO