摘要 |
PROBLEM TO BE SOLVED: To provide a voice recognizing device capable of satisfactorily preparing a dictionary for recognizing a specific speaker without casting a burden on the speaker. SOLUTION: This voice recognizing device has a standard dictionary 1, a feature extracting part 2 for extracting the quantity of features from inputted voice, a collating part 3 for collating the quantity of features extracted by the feature extracting part 2 with the standard dictionary 1, a result output part 4 for outputting results of the collation in the collating part 3, and a dictionary renewing part 5 for renewing the standard dictionary 1. The standard dictionary 1 is prepared, in its initial stage, as a dictionary for recognizing unspecific speakers, and the dictionary renewing part 5 is adapted to renew the standard dictionary 1 into a dictionary for recognizing a specific speaker based on results of the collation of the quantity of features extracted from inputted voice with the standard dictionary 1. |