Title:
音声認識装置
Document Type and Number:
Japanese Patent JP4565768
Kind Code:
B2
Abstract:
PROBLEM TO BE SOLVED: To decrease the user's input number of steps (number of times of key operations) in speech inputs, and to enable the user to input a correct command. SOLUTION: In the speech recognition method by which the probability of each candidate word to be a speech input word is calculated, and the candidate word whose probability is maximum is recognized to be the speech input word, (1) the maximum probability is compared with a setting probability, (2) a command according to the recognized word is carried out when the maximum probability is larger than the setting probability, (3) the setting probability is decreased when the cancellation of the command is not instructed, and (4) the setting probability is increased while canceling the performance of the command when the cancellation of the command is instructed.
Inventors:
Shuichi Matsumoto
Application Number:
JP2001120777A
Publication Date:
October 20, 2010
Filing Date:
April 19, 2001
Export Citation:
Assignee:
Alpine Electronics Co., Ltd.
International Classes:
G10L15/00; G10L15/08; G10L15/22; G10L15/28
Domestic Patent References:
JP11249688A | ||||
JP4152397A | ||||
JP4032900A | ||||
JP62032500A | ||||
JP59045499A |
Attorney, Agent or Firm:
Chimitsu Saito