| 摘要 |
PROBLEM TO BE SOLVED: To provide a voice recognition apparatus and a voice recognition program, capable of improving recognition performance without needing user's troublesome operation. SOLUTION: The apparatus comprises: a vocabulary dictionary storing section 11 in which a plurality of vocabularies are stored beforehand; a vocabulary dictionary managing section 12 for extracting the vocabulary to be recognized; a collation section 13 for calculating a matching rate with the vocabulary to be recognized, based on a received voice; a result output section 14 for outputting the vocabulary whose score is the best, from the calculation result of the matching rate, as the recognition result; and an extraction reference information managing section 12c for changing extraction reference information according to a monitor result by a monitor control section 15. The vocabulary dictionary storing section 11 further includes: a measure information storing section for storing measure information 20 which becomes a measure when the vocabulary to be recognized is extracted; and an extracting reference information storing section in which extracting reference information 24 for indicating reference of the measure information 20, when the vocabulary to be recognized is extracted, is stored. When the extracting reference information 24 is changed, the vocabulary dictionary managing section 12 increases or decreases the number of vocabularies to be recognized. COPYRIGHT: (C)2008,JPO&INPIT
|
