PURPOSE: To prevent an input voice from being misrecognized even at the time of inputting a voice overlapping a signal sound due to user's carelessness by sending vocalization timing correcting guidance at the time of detecting an overlap between a guidance signal and response vocalization.
CONSTITUTION: When a line is connected by a line control part 5 in the voice recognizing/answering device 3, a required sound to be guidance is synthesized at a voice synthesizing part 7 through a main control part 8 and a signal tone is sent to a user's telephone set 1. A user's vocalized sound responding to the guidance is set to a voice recognizing part 6, a voice part is segmented and the existence of an overlap between the signal tone and the vocalized sound is detected by an overlap detecting part. At the time of detecting the overlap, succeeding voice recognizing operation is stopped and the voice synthesizing part 7 synthesizes and sends a guidance synthetic sound indicating that the vocalized sound can not be confirmed due to the overlap to the signal tone based upon a command from the control part 8. Since the vocalization timing is corrected, the misrecognition of an input voice can be prevented even at the time of inputting a voice overlapped to a signal tone due to user's carelessness.
MIYAMAE TORU
YAHATA MOTOYUKI