| 摘要 |
PROBLEM TO BE SOLVED: To avoid repeating of wrong recognition of a voice recognition result. SOLUTION: A voice input from a voice input section 210 is recognized by an acoustic processing section 202 to a search section 204, and a plurality of recognition results are output together with recognition scores. A recognition result selection section 205 selects a prescribed number of recognition results in the decreasing order of recognition scores; and when a recognition result output section 209 outputs the prescribed number of selected recognition results, the recognition result of the highest ranking is held in a recognition result history 208. At this time, when a current recognition result in the top ranking matches the last history held in the recognition result history 208, the recognition result has its ranking lowered in the order and is output, thereby preventing the same misrecognition from being repeated. COPYRIGHT: (C)2007,JPO&INPIT
|
