Identification of steady-state vowels synthesized from the Peterson and Barney measurements

Abstract

The purpose of this study was to determine how well listeners can identify vowels based exclusively on static spectral cues. This was done by asking listeners to identify steady-state synthesized versions of 1520 vowels (76 talkers x 10 vowels x 2 repetitions) using Peterson and Barney's measured values of F0 and F1-F3 [J. Acoust. Soc. Am. 24, 175-184 (1952)]. The values for all control parameters remained constant throughout the 300-ms duration of each stimulus. A second set of 1520 signals was identical to these stimuli except that a falling pitch contour was used. The identification error rate for the flat-formant, flat-pitch signals was 27.3%, several times greater than the 5.6% error rate shown by Peterson and Barney's listeners. The introduction of a falling pitch contour resulted in a small but statistically reliable reduction in the error rate. The implications of these results for interpreting pattern recognition studies using the Peterson and Barney database are discussed. Results are also discussed in relation to the role of dynamic cues in vowel identification.

Pitch characteristics of infant-directed speech affect infants’ ability to discriminate vowels

“…But resonances can be detected only if there is energy in the signal close to the frequency of the resonance. The sound source for speech, the larynx, produces energy only at the frequencies of the fundamental and its harmonics, which are integer multiples of the fundamental.When there is a pitch contour present, the fundamental frequency changes, and the frequencies of the harmonics follow, whereas the frequencies of formant resonances remain relatively constant (Hillenbrand & Gayvert, 1993). Thus, when there is a large pitch change, it is likely that, at some point, the harmonics will sweep through the frequencies of the formants, thereby disclosing their locations (see Figure 1).…”

Perception of dynamic information for vowels in syllable onsets and offsets

“…Myron L. Braunstein sustained vowels produced by trained speakers and gated to produce 300-msec samples. More recently, Hillenbrand and Gayvert (1993) found that synthetic static vowels matched to the values ofthe Peterson and Barney stimuli were not well identified, although Peterson and Barney's listeners had high rates of identifiability for the original stimuli.…”

Formant detail needed for vowel identification

“…Results for each talker are shown in Tables 1 and 2. Identification scores for the natural vowels (89%) were lower than those from Peterson and Barney (1952), and accuracy levels for the 1-point, flat-formant vowels (64%) were lower than scores for the flat-formant vowels in Hillenbrand and Nearey (1999) and Hillenbrand and Gayvert (1993). The level of formant detail significantly influenced identification accuracy ͓F(5,50) ϭ123.0;pϽ0.01͔.…”

“…American English listeners use dynamic formant transitions to identify vowels in the absence of steady-state vowel target frequencies (e.g., Strange et al, 1983;Nearey and Assmann, 1986). In addition, listeners identify vowel stimuli containing time-varying formants better than stimuli with steady-state formants (Hillenbrand and Gayvert, 1993). Little is known, however, about how much formant contour detail is used in vowel perception.…”