| 摘要 |
PURPOSE:To obtain a natural synthesized voice by putting a 1st and a 2nd membership function together and calculating the continuance of a syllable, and finding the continuance of a phoneme according to the continuance of the syllable by using a table which contains continuance data on consonants and vowels in the syllable. CONSTITUTION:A rhythm processing part 23 finds the 1st membership function showing the appearance probability of continuance per mora corresponding to rhythmic environment and the 2nd membership function showing the appearance probability of the continuance of the syllable corresponding to phonemic environment from a memory 15 for rhythm control, and then calculates the continuance of the syllable by putting the 1st and 2nd membership functions together. Then the continuance of the phoneme is found based on the continuance of the syllable by using the table containing continuance data on the consonants and vowels in the syllable. Consequently, the synthesized voice with high naturalness which has continuance close to a natural voice can be obtained. |
