Identification of resynthesized /hVd/ utterances: Effects of formant contour

The purpose of this study was to examine the role of formant frequency move ments in vowel recognition. Measurements of vowel duration, fundamental fre quency, and formant contours were taken from a database of acoustic measure ments of 1668 /hVd/ utterances spoken by 45 men, 48 women, and 46 children [Hillenbrand etal., J. Acoust. Sec. Am. 97, 3099-3111 (1995)]. A 300-uttera nce subset was selected from this database, representing equal numbers of 1 2 vowels and approximately equal numbers of tokens produced by men, women, and children. Listeners were asked to identify the original, naturally prod uced signals and two formant-synthesized versions. One set of "original for mant" (OF) synthetic signals was generated using the measured formant conto urs, and a second set of ''flat formant'' (FF) signals was synthesized with formant frequencies fixed at the values measured at the steadiest portion of the vowel. Results included: (a) the OF synthetic signals were identifie d with substantially greater accuracy than the FF signals; and (b) the natu rally produced signals were identified with greater accuracy than the OF sy nthetic signals. Pattern recognition results showed that a simple approach to vowel specification based on duration, steady-state F-0, and formant fre quency measurements at 20% and 80% of vowel duration accounts for much but by no means all of the variation in Listeners' labeling of the three types of stimuli. (C) 1999 Acoustical Society of America. [S0001-4966(99)04406-9] .