摘要 |
PROBLEM TO BE SOLVED: To make it possible to set a phonological time length of a phonological series so that it becomes a specified utterance time and to provide a natural phonological time length regardless of the length of an utterance time. SOLUTION: An utterance time T of the phonological series is determined based on a control data stored at a step S2, and is defined as an utterance time of the phonological series produced at a step S3. A sum of the phonological times of the phonological series is set so as to become equal to the utterance time by repeating step S4 through step S8 and S10, and a speech waveform is produces by connecting speech elements based on the voice pitch and the phonological time length of the synthesized speech. Here, in the case of setting phonological time lengths of each phoneme, an initial value of each phonological time length is set (step S5) to be within the range of the phonological time length determined from an average value, the standard deviation, and the minimum value, and based on the initial value of this phonological time length, the phonological time lengths are set (step S6). |