MULTIMODAL SYSTEM AND SPEECH INPUT METHOD

摘要

PROBLEM TO BE SOLVED: To provide a multimodal system that enables a user to input data without remembering speech recognition command and can improve a speech recognition rate.

SOLUTION: A server 2 sends a candidate for an input character to a terminal 1, which displays the candidate for the input character sent from the server 2 on a screen. The user pronounces the candidate for the input character which is displayed on the screen to input a voice. The terminal 1 sends the voice to a voice portal 3 through a communication line. The voice portal 3 recognizes the voice and sends the recognition result to the server 2. Then the voice portal 3 has a database in which speaking frequencies by users are recorded and updates the database each time a voice is received from a user. The server 2 determines a candidate for an input character to be sent to the terminal 1 according to the database.

COPYRIGHT: (C)2005,JPO&NCIPI