摘要 |
PROBLEM TO BE SOLVED: To obtain an audio response apparatus which, when the representation of a synthetic voice is requested, detects the semantic break of a sentence by going back in a direction opposite to the direction of time from a point of time when it received the request and performs the representation by taking there as point a representation starting time of the synthetic voice. SOLUTION: A waveform editing part 3 calls out data corresponding to an input series 2 from a word voice database 4 and a template 5 to generate a synthetic voice 8 having the length of a sentence and the length of a document by connecting words in the database 4 and phrases in template 5. When a user askes the synthetic voice again, a command input 6 is imparted. When the equipment receives the command input 6, a rythm boundary detecting part 7 detects a rythm boundary by going back along the synthetic voice 8 in the direction opposite to the direction of time from a point of time when the equipment receives the input 6 and starts the presentation of the synthetic voice 8 from that place. |