The effect of interrelation between voice pitch and formant frequencies on the perception of synthetic vowels was investigated. Using a software cascade parallel formant speech synthesizer SMOK, modelled on the system of D. H. Klatt and capable of producing high quality speech, six Polish vowels /i/, /i/, /e/, /a/, /o/, /u/ spoken by male and female speakers were resynthesised. Those prototypical vowels were subsequently modified by varying fundamental frequency F0 in the range of ±1 octave in relation to the original values. 15 subject (10 male and 5 female) participated in the listening experiments. Their task was to identify vowel stimuli presented in random order. The results indicate that it is not formant frequencies alone, but also their relation to F0. That is of crucial importance in the perceptual identification of vowels. Discrepancy between formant frequencies and voice pith may lead to a shift in phonetic category.
Bibliographic reference. Owsianny, Mariusz (1995): "The effect of voice pitch on perception of synthetic Polish vowels", In EUROSPEECH-1995, 945-948.