Speech Pathol. and Audiol., Western Michigan Univ., Kalamazoo, MI 49008
The purpose of this study was to examine the role of formant frequency movements in vowel recognition. Measurements of fundamental frequency and formant contours were taken from a database of acoustic measurements of 1668 /hVd/ utterances spoken by 45 men, 48 women, and 46 children [Hillenbrand et al., J. Acoust. Soc. Am. 97 (in press)]. A total of 300 utterances were selected from this database, representing equal numbers of 12 vowels (i,(small capital eye),e,(cursive beta),(ae ligature),a,(open oh),o,(small capital you),u,(inverted vee),(hooked backward eh)) and approximately equal numbers of tokens produced by men, women, and children. Listeners were asked to identify three versions of each utterance: (1) the original, naturally produced signal, (2) a formant-synthesized version that followed the measured contours of F1--F3, and (3) a formant-synthesized version with F1--F3 fixed at their steady-state values. Preliminary results indicate: (1) the original signals are identified with substantially greater accuracy than either of the synthetic versions, and (2) the synthetic signals that follow the original formant contours are identified with greater accuracy than the flat-formant signals.