Speaker identification in hybrid synthesis: Implications for speech perception

Abstract

As part of an evaluation of hybrid synthesis [Hertz, Proc. IEEE Workshop on Speech Synthesis (2002)], perceptual experiments were conducted that tested the hypothesis that stressed vowels are the primary cues to speaker identification. Hybrid sentences were constructed for eight voices, including child and adult and male and female, in which stressed vowels were taken from a single human speaker, but other segments were replaced by surrogates from different sources. Some surrogates were natural speech segments; others were formant synthesized. Some matched the age or gender of the stressed vowel speaker; others did not. After being trained on six human target voices, listeners were asked to identify the hybrid stimuli, and also fully synthetic and natural stimuli (for target and nontarget voices), in terms of age, gender, and whether and how much they matched a target voice. For all categories, hybrid stimuli, in contrast to synthetic ones, were identified as accurately as natural speech, both by listeners familiar with a target voice (e.g., family members) and those unfamiliar. The experiment demonstrated how little attention is paid to consonants and reduced vowels in determining speaker identity, and supported our hybrid approach to modeling voices. [Work supported by NIH Grant 1 R43 DC006761-01.]