Perceptual evaluation of naturalness due to substitution of Chinese syllable for concatenative speech synthesis

摘要

In order to estimate degradation of naturalness in concatenative speech synthesis due to mismatch of syllables between unit selection and its use, perceptual experiments were conducted using speech stimuli synthesized by concatenating the final of first syllable and the initial of second syllable. The results for substitution of the succeeding final of syllable and preceding initial of syllable showed that the naturalness was low when the speech segmentation was difficult, like [s, m, y,.....]. For the substitution of the tone, the results showed that the naturalness was (1) high for first and second tone combination, (2) low for third and second tone combination.