PURPOSE: To reduce time and labor required for the verification of an error by synthesizing voices based on the reading of each code comprising a character string when the error occurs in the voice due to reading decided by using a dictionary.
CONSTITUTION: In the dictionary 111, the character string and the reading are stored as a pair, and the reading of the character string corresponding to an introduced character code string is decided and outputted by a first reading conversion means 121 based on the dictionary 111. Also, the reading of each code comprising an introduced character string is outputted by a second reading conversion means 131. A voice synthesis means 141 performs voice synthesis based on the reading outputted from either two reading conversion means 121 and 131. When the error occurs in the voice due to the reading decided by using the dictionary 111, the voices based on the reading of each code comprising the character code string are synthesized. In such a way, it is possible to easily verify whether the error exists in the decision of the reading by the first reading conversion means 121 or in an inputted character code string itself.
KASEDA MITSUKO
TERAMOTO YOSHIAKI