Dynamic Vowel Research Articles

VoxSenes Acoutic Database

Motivated by the need to increase knowledge about the effect of age on speech production, and by the lack of speech data on European Portuguese (EP), this work focuses on the development of a speech database, with acoustic data of healthy adults using standardized recording procedures. The objective of this database is to support a comprehensive analysis of age and gender effects in EP speech at segmental and suprasegmental levels. In this vein, the present work provides the description of one acoustic speech database containing all EP vowels produced in similar context (reading speech), and also semi-spontaneous speech (image description) collected from a large sample of 144 adults from the central region of Portugal, between the ages 35 and 97. This paper summarizes the research design of the acoustic database and includes details about the study region, participants, participants’ tasks, corpus, data acquisition, and segmentation/annotation of the data. This database has been used in different studies, where various aspects of the aging speech were explored through different methods, such as static acoustic features and dynamic vowel formants. However, the collected data could be used on other acoustic studies of speech over the lifespan.

Dialect levelling and Cockney diphthong shift reversal in South East England: the case of the Debden Estate

This article explores an instance of dialect levelling in South East England, the reversal of Cockney diphthong shift. We trace this reversal through an apparent-time analysis of 52 speakers from Debden, a community in Essex with East London heritage. Dynamic vowel analyses of word-list and passage data suggests a reversal of the diphthong shift towards SSBE targets which has occurred most abruptly in those born after 1992 potentially as a result of increased social mobility in this generation. We compare the results in Debden to previous findings in the south-eastern towns of Milton Keynes and Reading where apparent-time change was also observed away from a shifted vowel system and towards SSBE targets (Kerswill & Williams 2000, 2005). In diverse areas of South East England, a common process of levelling towards the pan-regional standard is present which is not occurring exclusively as a result of dialect contact or face-to-face interaction. Nonetheless, each community exhibits a distinct trajectory and timing of language change which can be attributed to different patterns of movement and resettlement and, in particular, access to social mobility and the retention of community networks.

Dynamic targets in second language vowel articulations

This study examines L1 English-L2 French vowel productions, with the goal of exploring whether representing vowel targets as dynamic rather than “steady-state” better predict L2 production patterns. It is hypothesized that learners will transfer L1 vowel dynamic information to produce L2 vowels if the two are perceived as “similar,” but it is unclear whether learners will transfer dynamic vowel information to “new” L2. As such, six L1 English-L2 French learners completed production tasks in English and French with the target vowels /i, u, e, o/ and /y, ø/ in real words. Participants were recorded using a stabilization headset attached to an ultrasound probe and microphone. F1 and F2 measurements and tongue contours were extracted at three timepoints throughout the vowel duration. Results shows that while learners produce L2 French vowels /i, e, u, o/ with a similar amount of formant and gestural movement to L1 vowels, they do not transfer midpoint F1-F2 values. Learners do produce new L2 vowels /y, ø/ with similar formant and gestural movements to English vowels. The results of this study indicate that learners transfer dynamic acoustic information and articulatory gestures from their L1 to produce L2 vowels, which has implications for the representation of vowels.

Visual perception of vowels from static and dynamic cues.

The purpose of the study was to analyse human identification of Polish vowels from static and dynamic durationally slowed visual cues. A total of 152 participants identified 6 Polish vowels produced by 4 speakers from static (still images) and dynamic (videos) cues. The results show that 59% of static vowels and 63% of dynamic vowels were successfully identified. There was a strong confusion between vowels within front, central, and back classes. Finally, correct identification strongly depended on speakers, showing that speakers differ significantly in how "clearly" they produce vowel configurations.

Finite Element Synthesis of Diphthongs Using Tuned Two-Dimensional Vocal Tracts

Three-dimensional 3-D vocal tract acoustic modeling has the potential to generate high quality and natural voice sounds, but at the price of a large computational cost. Alternatively, 2-D models based on tuned vocal tracts have shown to provide similar results to the 3-D ones but with less computational demands. However, they are currently limited to the synthesis of static vowel sounds. In this paper, the tuned 2-D approach is extended by considering moving vocal tracts to generate dynamic vowel sounds, like diphthongs. Four tuning steps are followed to build a dynamic 2-D vocal tract model that can recover, to a large extent, the formant locations, bandwidths, and energies of a 3-D vocal tract with circular cross section, set in a spherical baffle representing the human head. Acoustic waves propagating through the time evolving vocal tract and radiating to free-field are simulated using the finite element method in the time-domain. As examples, the diphthongs [Ai] and [Au] have been generated using the tuning approach and compared, by means of objective and subjective evaluations, to those resulting from 3-D and conventional 2-D simulations.

More on Michigan Listeners' Perceptions of /ɑ/-Fronting

The Northern Cities Vowel Shift (NCVS) is well advanced in southeastern Lower Michigan and largely absent in Michigan's Upper Peninsula. In a previous study, Plichta and Rakerd found evidence that listeners from Lower Michigan make a perceptual adjustment for NCVS /ɑ/-fronting when given acoustical information about a talker's dialect (a sentence-long speech sample), but listeners from the Upper Peninsula do not. The present study further examines this phenomenon. Here, individual differences regarding listeners' vowel categorization judgments indicate that the /ɑ/-fronting perceptual response is consistently present among young adult residents of Lower Michigan, but much more variable among young adults from the Upper Peninsula. There is also evidence that the perceptual effect for Lower Michigan listeners is robust over a number of different stimulus variations. A word-level difference found here suggests that perceptual cues to /ɑ/-fronting may be provided by both static and dynamic vowel formant features. Finally, listeners' decision reaction times provide new evidence of regional differences in sensitivity to NCVS.

Monophthongs and formant movement in North American English

This poster describes dynamism in vowel production using data derived from two experimental studies, one in Southern California, the other in southern Manitoba. In both studies, young adults read from a script containing multiple repetitions of target vowels in /hVd/ context. The frequencies of the first four formants were measured at three time points (25%, 50%, 75% of vowel duration) in each vowel. Analyses were conducted in both linear (hertz) and auditorily transformed (bark distance) terms. Many dialects of English clearly have diphthongal production of the tense vowels, especially /e, o/, which would be indicated by movement of the formants toward the periphery of the vowel space. For the two populations studied here, none of the tense vowels /i, e, o, u/ show net movement toward the periphery. While there are individual and group differences, simplex vowels in both dialects show very little movement between time points one and two, and only non-targeted, transitional movement between time points two and three. This suggests monophthongal basic representations, though not ‘‘flat’’ or steady productions, for these two dialects. Implications for North American dialectology and for dynamic vowel models will be considered.

Categorization and discrimination of nonspeech sounds: differences between steady-state and rapidly-changing acoustic cues.

Different patterns of performance across vowels and consonants in tests of categorization and discrimination indicate that vowels tend to be perceived more continuously, or less categorically, than consonants. The present experiments examined whether analogous differences in perception would arise in nonspeech sounds that share critical transient acoustic cues of consonants and steady-state spectral cues of simplified synthetic vowels. Listeners were trained to categorize novel nonspeech sounds varying along a continuum defined by a steady-state cue, a rapidly-changing cue, or both cues. Listeners' categorization of stimuli varying on the rapidly changing cue showed a sharp category boundary and posttraining discrimination was well predicted from the assumption of categorical perception. Listeners more accurately discriminated but less accurately categorized steady-state nonspeech stimuli. When listeners categorized stimuli defined by both rapidly-changing and steady-state cues, discrimination performance was accurate and the categorization function exhibited a sharp boundary. These data are similar to those found in experiments with dynamic vowels, which are defined by both steady-state and rapidly-changing acoustic cues. A general account for the speech and nonspeech patterns is proposed based on the supposition that the perceptual trace of rapidly-changing sounds decays faster than the trace of steady-state sounds.

Cue weighting of static and dynamic vowel properties in children versus adults

The purpose of this study was to determine whether children give more perceptual weight than do adults to dynamic spectral cues versus static cues, when identifying vowel sounds. Three experimental stimulus sets were presented, each with 30-ms stimuli. The first consisted of unchanging formant onset frequencies ranging in value from frequencies for [i] to those for [ɑ], corresponding to a bilabial stop consonant. The second two consisted of either an [i] or [ɑ] onset frequency with a 25-ms portion of a formant transition whose trajectory was toward one of a series of target frequencies ranging from those for [i] to those for [ɑ]. Ten children between the ages of 3;8 and 4;1 and a control group of 10 adults identified each stimulus as [bi] or [bɑ]. The results showed developmental effects: the children relied more heavily than the adults did on the static formant onset frequency cue to identify the vowels, while the adults appeared to give more equal weight to both static and dynamic cues than the children did. These findings contradict the Developmental Perceptual Weighting Shift theory and are discussed in relation to this theory and other current research on the development of vowel perception.

Spectral tilt versus formant frequency in static and dynamic vowels

The relative importance of gross (e.g., spectral tilt) versus detailed (e.g., formant peak frequencies) for speech perception is an enduring question of substantial importance in understanding performance of both impaired and unimpaired listeners [cf. Kiefte, J. Acoust. Soc. Am. (submitted)]. In this study, subjects were presented with steady-state stimuli varying perceptually between [i] and [u] for which spectral tilt and formant frequencies were fully crossed. Results show an effect for both tilt and formants indicating some perceptual importance for both in the identification of statically specified vowels. However, this result does not appear to extend to more natural sounding stimuli with dynamic formants. For example, there was no effect for spectral tilt in a similar experiment using diphthongs ranging from [ai] to [au]. One hypothesis for this difference is that the F2 of front vowels is masked by F1 when combined with sharply falling spectral tilt and that this effect is mitigated in dynamic or naturally produced vowels. Results are also presented which compare the perception of these continua with similar stimuli in which F2 is absent as well as with dichotically presented stimuli in which F1 and F2 are heard in different ears. [Work supported by NIDCD DC04072.]

Static and dynamic vowels in a "cepstro-phonetic" sub-space.

Static and dynamic vowels in a "cepstro-phonetic" sub-space.

Investigating the ‘‘trough:’’ Vowel dynamics and aerodynamics

Articulatory coordination in vowel–consonant–vowel utterances (V=/i a/, C=/p b/) was studied in 5 Swedish, 3 English, 2 German, and 3 French speakers. Electropalatographic records showed that the amount of transconsonantal V-to-V coarticulation was greater in German and French than in Swedish and English. For the symmetrical utterances /ipi/ and /ibi/, the EPG patterns displayed a marked ‘‘trough’’ (a temporary reduction of tongue height) associated with the stop closure in the Swedish and English, but not in the German and French speakers; the Swedish and English trough patterns were, however, differently timed. These observations are consistent with dynamic vowel features characteristic of the respective languages. In a second experiment, using a wider voice onset time range, three Swedish speakers produced words containing CV syllables (C=/p b t d k g/, V=/i o/ a ■ u/). VOT was controlled in terms of voicing, stress, word length, and position in the word. Formant data showed that the amount of F2 locus-to-target assimilation decreased as a function of voicing lag. This suggests that aerodynamic factors associated with aspiration affect anticipatory coarticulation patterns; specifically, it may affect trough patterns in symmetrical VCVs.

Implicit memory for hybrid silent-center syllables

A growing body of evidence suggests that information about specific voices and specific exemplars of words is encoded in memory, as observed in results of implicit memory tasks in which subjects perform perceptual classification of stimuli. Hura [dissertation, University of Texas (1994)] found identification performance on silent-center (SC) syllables improved with previous exposure to corresponding full syllables, implying that representations of individual full syllables were encoded in memory and accessed in identification of SC stimuli. The current study investigates whether prior exposure to SC syllables improves performance on full syllables. Ten /bVt/ syllables spoken by multiple talkers were used to generate SC and hybrid SC stimuli (which combine the initial CV transition from one talker and the final VC transition from another). A subset of SC syllables, varying for identity of talker or vowels presented, were presented to subjects for identification. After a variable delay subjects were tested on a subset of full syllable stimuli. The magnitude of implicit memory effects in each case is assessed to address the claim that dynamic vowel information is speaker independent as suggested by the theory of dynamic specification.

Speaker normalization of static and dynamic vowel spectral features

Two methods are described for speaker normalizing vowel spectral features: one is a multivariable linear transformation of the features and the other is a polynomial warping of the frequency scale. Both normalization algorithms minimize the mean-square error between the transformed data of each speaker and vowel target values obtained from a "typical speaker." These normalization techniques were evaluated both for formants and a form of cepstral coefficients (DCTCs) as spectral parameters, for both static and dynamic features, and with and without fundamental frequency (F0) as an additional feature. The normalizations were tested with a series of automatic classification experiments for vowels. For all conditions, automatic vowel classification rates increased for speaker-normalized data compared to rates obtained for nonnormalized parameters. Typical classification rates for vowel test data for nonnormalized and normalized features respectively are as follows: static formants--69%/79%; formant trajectories--76%/84%; static DCTCs 75%/84%; DCTC trajectories--84%/91%. The linear transformation methods increased the classification rates slightly more than the polynomial frequency warping. The addition of F0 improved the automatic recognition results for nonnormalized vowel spectral features as much as 5.8%. However, the addition of F0 to speaker-normalized spectral features resulted in much smaller increases in automatic recognition rates.

The roles of three possible cues to dynamic vowel identity

Recent research [Nearey and Assmann (1986);Strange (1989)] suggests that the changes in vowels over time are perceptually relevant to their identification. Possible “cues” for dynamic vowels include the frequency locus of spectral peaks over time, duration, and extent, rate, and direction of peak movement over time. The possible roles of duration and extent and rate of peak movement over time in cuing the [ʌ]‐[a] distinction were examined using synthetic [ʌ]‐[a] series. For each of three continua, two of the cues were varied while the third was held constant. Pilot testing showed that the continua were ambiguous on the basis of frequency alone and direction of spectral peak movement over time was nondistinctive. The ability of subjects to use the two vowel categories consistently was the basis on which the efficacy of the cues was judged. Duration was found to reliably cue vowel identity while no evidence of an effect of either rate of movement of spectral peak change or extent of spectral peak movement was found. Results will be discussed with respect to potential acoustic “cues” for dynamic vowels. [Work supported by NIDCD Grant DC 00219 to SUNY at Buffalo.]

A Perceptual Mechanism for Dynamic Vowels Observed by Experiments of Dichotic Fusion of Synthetic Speech Sounds

A Perceptual Mechanism for Dynamic Vowels Observed by Experiments of Dichotic Fusion of Synthetic Speech Sounds