| 摘要 |
<p>PURPOSE: To provide a voice input method and a device therefor in which the previous recognition result, a likelihood, and the next recognition result are used to judge whether the previous recognition result is right or not to eliminate the transmission of information by erroneous recognition, and the time necessary for the transmission of information is shortened. CONSTITUTION: This device has a voice input part 1 for inputting the voice of a speaker, a voice recognizing part 2 for extracting the characteristic pattern from the input voice performing a pattern matching between the characteristic pattern and a builtin voice pattern, and outputting the recognition result and the tolerance, a recognition result output part 4 for showing the recognition result to the speaker, a likelihood judging part 6 for judging the propriety of the previously inputted recognition result on the basis of the previously inputted recognition result, the likelihood and the input result inputted next, and outputting the likelihood final recognition result, and an information transmitting part 6 for outputting the information to the input of the final recognition result.</p> |
