Title:
【発明の名称】音声認識装置
Document Type and Number:
Japanese Patent JP2836102
Kind Code:
B2
Abstract:
PURPOSE:To accurately recognize the voice of an object person for voice recognition by comparing the voice with previously registered voice patterns by a voice pattern comparison part and further inferring whether the voice pattern is the voice pattern of the object body or not through the fuzzy inferring part according to its voice symptoms. CONSTITUTION:A memory part 2 is stored with the pattern of the input voice of the object person for voice recognition and a memory file part 3 is stored previously with a registered file of the voice pattern of the object part of voice recognition. Then the voice pattern comparison part 4 compares the input voice pattern from the memory part 2 with the registered voice pattern from the memory file part 3 and outputs the degree of the coincidence between both voice patterns. The voice symptoms from a voice symptom selecting operation part 5, on the other hand, are outputted to the fuzzy inferring part 6, which infers whether the input voice is the voice of the object person of voice recognition or not according to a fuzzy rule by using the degree of coincidence between both voice patterns and the voice symptoms which are inputted. Consequently, the voice can accurately be recognized even if the conditions, etc., of the object person of voice recognition are not good.
Inventors:
OKAMOTO KYOHIDE
Application Number:
JP16054589A
Publication Date:
December 14, 1998
Filing Date:
June 22, 1989
Export Citation:
Assignee:
OMURON KK
International Classes:
G10L15/24; G06F9/44; G06N7/02; G10L15/08; G10L15/10; (IPC1-7): G10L3/00; G06F9/44; G10L3/00
Domestic Patent References:
| JP272397A |
Other References:
【文献】IEEE Transactions on System,Man and Cybernetics,Vol.SMC-7,No.8,p.625~629(1977)
Attorney, Agent or Firm:
Kazuhide Okada