摘要 |
<p>PROBLEM TO BE SOLVED: To convert voice whose phoneme duration significantly extended so as to be heard as a more natural voice when the phoneme and particularly make a synthesized singing voice a voice including a vibrato close to a natural voice. SOLUTION: In a voice synthesizing device 10, a synthesis parameter generation part 14 performs control of reciprocally repeating the reading of a voice element successively read from an element dictionary 16 according to the voice duration within the range of a plurality of voice element and/or within a prescribed range around a basic frequency, and assigns it as a voice element piece and/or frequency (or pitch) corresponding to the read obtained by analyzing a text to generate a synthesis parameter. According to this, voice fluctuated according to the synthesis parameter is outputted during the voice duration.</p> |