| 摘要 |
<P>PROBLEM TO BE SOLVED: To generate script for voice synthesis for synthesizing voices of high quality in a small amount of data. <P>SOLUTION: When input data 1 such as a character string to be synthesized, a segment position and a pitch string are input, a pitch constant section detecting part 2 detects a pitch constant section continuing a value in a certain threshold value within the pitch of the input data, and a vibrato section detecting part 3 detects a vibrato section in which the pitch oscillates. A musical scale extracting part 4 makes a longer section from among the pitch constant section and the vibrato section a reference pitch section, and outputs a musical scale corresponding to an average pitch in the section as the reference pitch. A synthesizing script preparing part 5 prepares the script for voice synthesis expressing pitch transition by high and low accent marks expressing relative change from the reference pitch. A voice synthesizer generates and synthesizes the pitch string from the high and low accent marks of the script for the voice synthesis. <P>COPYRIGHT: (C)2006,JPO&NCIPI |
