CHARACTER CONVERSION DEVICE FOR VOICE RECOGNITION AND METHOD THEREOF

摘要

PURPOSE: A character conversion device for voice recognition and a method thereof are provided to automatically convert characters inputted for the registration of voice recognition into pronunciation variants. The device and method are appropriate for converting text type character information into characters for voice recognition, without additional training for recognizing a variable vocabulary. CONSTITUTION: An input block(102) inputs character data for registration. A character and numerical DB(106) stores pronunciation variant data for characters that are registration objects and other types of characters, and pronunciation variant data for numbers. A first conversion block(104) searches the character and numerical DB(106), to convert characters that are registration objects and other types of character or numbers among the inputted character data into pronunciation variants. An exceptional word DB(110) stores pronunciation variant data for words exceptional to regular inflection. A second conversion block(108) searches the exceptional word DB(110), to convert exceptional words among the inputted characters into pronunciation variants. And a third conversion block(112) converts inputted characters into pronunciation variants for voice recognition according to a prescribed phonemic inflection rule, for the registration in a character DB(114) for voice recognition.