| 摘要 |
<p>PROBLEM TO BE SOLVED: To prevent a secret symbol sequence (password code) form being known by others even when the voice is heard by the others on recorded as to a method and a device for speaker authentication, which input the secret symbol sequence in voice and decide whether a speaker is the user oneself according to the result of voice recognition. SOLUTION: Kinds of symbols and kinds of colors are made to correspond to each other at random, the respective symbols are displayed in the corresponding colors, and the user is instructed to input respective symbols in the secret symbol sequence and the displayed color names in voice; and the color names that the user voices are sequentially recognized and converted into a symbol sequence, which is compare with the previously registered symbol sequence. When they match each other, it is decided that the user is the subject.</p> |
