To improve the precision of speech synthesis while suppressing storage capacity.
If a speech synthesizing terminal 1 causes a read error of document data when a language processing part 12 performs conversion into phonetic data by reference to a terminal word dictionary D1, a read error detection part 15 transmits the document data where the read error occurs to a speech synthesis managing server 2. The speech synthesis managing server 2 when receiving the document data where the read error occurs from the speech synthesizing terminal 1 performs language processing by using the same terminal word dictionary D1 as the word dictionary that the speech synthesizing terminal 1 has and also performs language processing by using a large-scale server word dictionary D3, and compares the result of the language processing using the terminal word dictionary D1 and the result of the language processing using the server word dictionary D3 with each other to transmits update contents of the terminal word dictionary D1 needed to eliminate the read error as update word data to the speech synthesizing terminal 1. The speech synthesizing terminal 1 updates the terminal word dictionary D1 by using the received update word data.
YAMAZAKI YASUSHI
KATAE NOBUYUKI
