| 摘要 |
PROBLEM TO BE SOLVED: To shorten a time necessary for recognition and to reduce recognition error by outputting, as a recognition result, an object word most similar to an input speech of generated words to be recognized based on features of the externally inputted speech. SOLUTION: This speech recognition device is made up with an analyzing part 11, a feature extracting part 12, a word collating part 17, a judging part 18, and a speech synthesizing part 19. An external input speech is received by using a microphone 10 and the input speech is analyzed by the analyzing part 11. And, the feature extracting part 12 extracts the feature from a result of the speech analysis, the word collating part 17 collates the feature, and it is judged whether or not the result of the collation coincides with the word of the order for this time. When the judging part 18 judges that the result of the collation coincides with the word of the order for this time, namely, an order data to be inputted, the speech synthesizing part 19 synthesizes a 'correct answer' of speech. On the other hand, when it is judged that the result of the collation does not coincide with the order data to be inputted now, the speech synthesizing part 19 synthesizes 'incorrect answer' of speech. |
