| 摘要 |
PROBLEM TO BE SOLVED: To report a lot of routine data in voice in a short time without getting tired by storing plural model sentences for embedding the routine data, and synthesizing sentences while changing the model sentence to be embedded for each piece of routine data. SOLUTION: A model sentence selecting part 1 inputs routine data from a routine data storage part 1-1 and selects a model sentence according to a predetermined prescribed selection method. Plural model sentences are stored in a model sentence storage part 1-2 while making different contents in the fixed character parts of sentences so as to embed all the respective routine data. In the case of reading routine data expressed as a table or the like, the sentence in the same format is not repeatedly read but different sentences can be read for each piece of routine data, overlapped data can be simultaneously read or only required information can be first heard. Therefore, a listener can listen to the required information in a short time without getting tired. |
