| 摘要 |
<p>PURPOSE: To perform natural and smooth interaction between a human and a computer by reacting to the key word of voice uttered by the human and inserting an agreeing response in a speech. CONSTITUTION: The voice inputted from a voice input part 1 is converted to a feature amount in an acoustic analysis part 2. In a matching part 3, the feature amount of input voice is compared with the feature amount of the key word registered in a key word storage part 4 beforehand and the key word is detected. At the time, the end time of the uttering of the key word is obtained by a time information obtaining part 5 and sent to an output part 6 along with the information of the key word. In the output part, the present time and the end time of the key word obtained from the time information obtaining part are compared, and when the difference becomes larger than a certain threshold value, the agreeing response is outputted.</p> |
