摘要 |
PROBLEM TO BE SOLVED: To efficiently understand utterance contents of a user by asking such a proper question that a reply from a user can precisely be recognized when interactively understanding the utterance contents of the user. SOLUTION: Various names and classification words representing attribute of the names are related to each other and stored as words to be recognized in an external storage device 7. Then the user utters a name and a signal processor 6 performs speech recognition processing for the utterance; when sufficient recognition likelihood can not be obtained, a question asking a reply by a classification word which is easy to recognize among classification words corresponding to the name is generated and outputted. COPYRIGHT: (C)2006,JPO&NCIPI
|