| 摘要 |
<P>PROBLEM TO BE SOLVED: To allow conversation to be smoothly developed between two speakers even when one speaker has been one-sidedly talking about a certain topic, by outputting, to the other speaker, specific sentences such as a sentence for making him or her participate the present conversation. <P>SOLUTION: This reply sentence retrieval system is provided with: a position presuming part 120 for presuming the position of each speaker based on the voice uttered by the speaker; a measurement part 130 for measuring each time interval, for each position, of the voice uttered from the speakers scattering in the respective positions based on the respective positions presumed by the presuming part 120; a ratio calculation part 150 for calculating the ratio of the time interval to a predetermined reference time for every time interval based on the respective time intervals measured by the measurement part 130; and a reply sentence retrieval part 160 for selecting one of the ratios from the respective ratios according to the values of the respective ratios calculated by the calculation part 150 to retrieve the reply sentence related to a code corresponding to the selected ratio. <P>COPYRIGHT: (C)2004,JPO |
