Spectral-shape features versus formants as acoustic correlates for vowels

Abstract

The first three formants, i.e., the first three spectral prominences of the short-time magnitude spectra, have been the most commonly used acoustic cues for vowels ever since the work of Peterson and Barney [J. Acoust. Soc. Am. 24, 175-184 (1952)]. However, spectral shape features, which encode the global smoothed spectrum, provide a more complete spectral description, and therefore might be even better acoustic correlates for vowels. In this study automatic vowel classification experiments were used to compare formants and spectral-shape features for monopthongal vowels spoken in the context of isolated CVC words, under a variety of conditions. The roles of static and time-varying information for vowel discrimination were also compared. Spectral shape was encoded using the coefficients in a cosine expansion of the nonlinearly scaled magnitude spectrum. Under almost all conditions investigated, in the absence of fundamental frequency (F0) information, automatic vowel classification based on spectral-shape features was superior to that based on formants. If F0 was used as an additional feature, vowel classification based on spectral shape features was still superior to that based on formants, but the differences between the two feature sets were reduced. It was also found that the error pattern of perceptual confusions was more closely correlated with errors in automatic classification obtained from spectral-shape features than with classification errors from formants. Therefore it is concluded that spectral-shape features are a more complete set of acoustic correlates for vowel identity than are formants. In comparing static and time-varying features, static features were the most important for vowel discrimination, but feature trajectories were valuable secondary sources of information.

The role of f 0 and formant frequencies in distinguishing the voices of men and women

“…Briefly, these problems include the following: (1) the determinacy problem, as Bladon (1982) has called it, which is the commonplace idea that tracking formants in natural speech is a difficult and, as yet, unresolved problem; (2) the straightforward observation that perceptual confusions made by human listeners nearly always involve speech sounds that are phonetically quite similar, a pattern that is difficult to reconcile with an underlying formant tracking process that is susceptible to gross errors that occur when formants either split or merge 4 (Klatt, 1982; see also Ito, Tsuchida, & Yano, 2001); (3) evidence showing that spectral details other than formant frequencies can affect phonetic quality (e.g., Bladon, 1982;Chistovich & Lublinskaja, 1979;Hillenbrand & Nearey, 1999). Partially in response to these problems, a number of investigators have argued that phonetic recognition is mediated by mental computations of similarities and differences in the gross shape of the spectrum rather than by formant frequencies (e.g., Bladon & Lindblom, 1981;Hillenbrand & Houde, 2003;Zahorian & Jagharghi, 1993). The issues discussed above revolve around the question of formants versus gross spectral shape in the perception of phonetic quality rather than talker sex, but the same question can be asked about either issue.…”

Temporal properties of perceptual calibration to local and broad spectral characteristics of a listening context

“…Again, while the role of spectral tilt in phonetic perception is unresolved (Hillenbrand et al, 1995;Bladon and Lindblom, 1981;Zahorian and Jagharghi, 1993;Ito et al, 2001), changes in tilt between speech segments have been shown to greatly influence speech perception by normal-hearing listeners (Alexander and Kluender, 2008), and especially hearing-impaired listeners (Alexander and Kluender, 2009) for whom spectral peaks are often reduced by abnormal cochlear filtering. The potential role that global spectral properties plays in speech perception is also demonstrated here with F 2 -matched precursors.…”

“…While the extent to which absolute spectral tilt plays a role in phonetic perception is generally unresolved (see, e.g., Hillenbrand et al, 1995;Bladon and Lindblom, 1981;Zahorian and Jagharghi, 1993;Ito et al, 2001), changes in tilt across time clearly influence perception of speech by normal-hearing listeners (Alexander and Kluender, 2008), and especially by hearing-impaired listeners (Alexander and Kluender, 2009) for whom spectral peaks are often obscured by abnormal cochlear processing. Furthermore, it is known that rapid changes in spectral tilt can substantially impair sentence identification (Van Dijkhuizen et al, 1987Haggard et al, 1987).…”

Cross-dialectal variation in formant dynamics of American English vowels

“…Having established the types of variation in formant trajectories that can be expected in cross-dialectal data in terms of TL, directionality, and curvature, a refinement of the current measures will be undertaken in order to address the changes in the direction of formant movement. In particular, parametrization procedures can be used ͑e.g., Harrington, 2006;Harrington et al, 2008;Hillenbrand et al, 2001;Morrison, 2009;Zahorian and Jagharghi, 1993͒ in order to model the various trajectory shapes.…”