| 摘要 |
<p>PROBLEM TO BE SOLVED: To provide a voice recognizing device with which recognition accuracy is improved by registering a recognized word after removing the personality of a speaker and normalizing the speaker when registering the word in a recognition dictionary storage part. SOLUTION: This device has a speaker normalizing part for normalizing the speaker personality of an audio signal pronounced by the speaker and outputting a speaker-normalized audio feature quantity, a voice recognition standard pattern storage part for storing a voice recognition standard pattern to be used for voice recognition, a syllable sequence recognizing part for recognizing the inputted audio feature quantity by using the voice recognition standard pattern, converting the quantity to syllable information and outputting the clause information, a recognition dictionary storage part for storing the text information of a recognition object and the syllable information corresponding to the text information, and a voice recognizing part, which has a dictionary register part for registering the text information and the syllable information in the recognition dictionary storage part, for recognizing the voice of an inputted audio signal when the audio signal is inputted, and outputting the text information of the recognized word as the recognized result.</p> |
