| 摘要 |
<P>PROBLEM TO BE SOLVED: To trace back and listen to a result of speech synthesis, in which a text is converted into a speech signal to be outputted, for only an arbitrary time. <P>SOLUTION: A text speech synthesis part 2 generates a speech signal according to input text information from an input text information part 1 and the speech signal is outputted from a speech output part 3. An output storage part 4 stores the speech signal up to the point of time traced back by a specified time length from the current point. A read-aloud position control part 6 traces back and read out of the output storage part 4 the speech signal by a time length corresponding to an indication supplied from a user to a read-aloud position indication part 5 and outputs the speech signal to a speech output part 3 to be reproduced. Thus, the speech signal which is generated and read aloud with a synthesized voice is traced back by the arbitrary time length and reproduced. At this time, the speech signal which is already generated is read aloud again, so the reproduction is carried out with the same reading, accent and intonation with the reproduction of the whole original sentence even when the sentence is reproduced from halfway. <P>COPYRIGHT: (C)2004,JPO |
