摘要 |
PROBLEM TO BE SOLVED: To provide a voice dialog system for an improved recognition rate of corrective speech and to provide a corrective speech recognition method. SOLUTION: Whenever a user depresses a correction switch 4b, a signal processor 14 makes an acoustic condition such as an acoustic model, a pose time, and an input gain a slow speech speed to normal speech, and modifies so as to correspond to speech voice input as corrective speech by punctuating one sound by one sound. COPYRIGHT: (C)2007,JPO&INPIT
|