| 摘要 |
<p>PROBLEM TO BE SOLVED: To provide a speaker collating device and method capable of collating definitely a speaker even when a synthetic voice expressing the personality of a speaker is used. SOLUTION: This method is a speaker collating method judging whether the voice of a speaker which is inputted coincides with the voice of a registered speaker which is preliminarily registered. In the method, an instruction is given to a speaker as to contents of uttered voices to be inputted and one or two or more voices of the speaker are inputted and the inputted voices are analyzed and degrees of similarity among two or more voices having the same uttered contents are calculated and, also, when they are degrees of similarity being equal to or higher than a prescribed level in which they are near to be completely coincident, the collating of the speaker is performed by using information that the degrees of similarity are coincident also.</p> |
