| 摘要 |
PURPOSE:To improve clearness by repeating a frame uniformly when the continu ation length is made long for a vowel or consonant part, setting a time length control section which controls the continuation length by thinning out frames when the continuation time is made short, and also setting a time length non- control section wherein neither the thinning nor repetition is performed for a voiceless plosive sound. CONSTITUTION:When the time length control section is set for the vowel or consonant part of 1-pitch data constitution and the continuation time length of the section is made long or short, frames are repeated or thinned out to facilitate the time length control and the naturalness of a synthesized voice is never spoiled. Further, when frames repeated or thinned out for a voiceless plosive sound, the clearness decreases, so neither the repetition nor the thinning is performed for the voiceless plosive sound. Consequently, a synthesized voice which is extremely close to a voice that a person speaks at the time of time length adjustment is obtained and the clearness is improved. |
