Status Of Vowels Research Articles

Optimization-based modeling of Lombard speech articulation: Supraglottal characteristics.

This paper shows that a highly simplified model of speech production based on the optimization of articulatory effort versus intelligibility can account for some observed articulatory consequences of signal-to-noise ratio. Simulations of static vowels in the presence of various background noise levels show that the model predicts articulatory and acoustic modifications of the type observed in Lombard speech. These features were obtained only when the constraint applied to articulatory effort decreases as the level of background noise increases. These results support the hypothesis that Lombard speech is listener oriented and speakers adapt their articulation in noisy environments.

Simulation of vowel-vowel utterances using a 3D biomechanical-acoustic model.

A link is established between biomechanical and acoustic 3D models for the numerical simulation of vowel-vowel utterances. The former rely on the activation and contraction of relevant muscles for voice production, which displace and distort speech organs. However, biomechanical models do not provide a closed computational domain of the 3D vocal tract airway where to simulate sound wave propagation. An algorithm is thus proposed to extract the vocal tract boundary from the surrounding anatomical structures at each time step of the transition between vowels. The resulting 3D geometries are fed into a 3D finite element acoustic model that solves the mixed wave equation for the acoustic pressure and particle velocity. An arbitrary Lagrangian-Eulerian framework is considered to account for the evolving vocal tract. Examples include six static vowels and three dynamic vowel-vowel utterances. Plausible muscle activation patterns are first determined for the static vowel sounds following an inverse method. Dynamic utterances are then generated by linearly interpolating the muscle activation of the static vowels. Results exhibit nonlinear trajectory of the vocal tract geometry, similar to that observed in electromagnetic midsagittal articulography. Clear differences are appreciated when comparing the generated sound with that obtained from direct linear interpolation of the vocal tract geometry. That is, interpolation between the starting and ending vocal tract geometries of an utterance, without resorting to any biomechanical model.

MRI-Based Vocal Tract Representations for the Three-Dimensional Finite Element Synthesis of Diphthongs

The synthesis of diphthongs in three-dimensions 3D involves the simulation of acoustic waves propagating through a complex 3D vocal tract geometry that deforms over time. Accurate 3D vocal tract geometries can be extracted from Magnetic Resonance Imaging MRI, but due to long acquisition times, only static sounds can be currently studied with an adequate spatial resolution. In this work, 3D dynamic vocal tract representations are built to generate diphthongs, based on a set of cross-sections extracted from MRI-based vocal tract geometries of static vowel sounds. A diphthong can then be easily generated by interpolating the location, orientation and shape of these cross-sections, thus avoiding the interpolation of full 3D geometries. Two options are explored to extract the cross-sections. The first one is based on an adaptive grid AG, which extracts the cross-sections perpendicular to the vocal tract midline, whereas the second one resorts to a semi-polar grid SPG strategy, which fixes the cross-section orientations. The finite element method FEM has been used to solve the mixed wave equation and synthesize diphthongs [${\alpha i}$] and [${\alpha u}$] in the dynamic 3D vocal tracts. The outputs from a 1D acoustic model based on the Transfer Matrix Method have also been included for comparison. The results show that the SPG and AG provide very close solutions in 3D, whereas significant differences are observed when using them in 1D. The SPG dynamic vocal tract representation is recommended for 3D simulations because it helps to prevent the collision of adjacent cross-sections.

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.

Extension of the single-matrix formulation of the vocal tract: Consideration of bilateral channels and connection of self-oscillating models of the vocal folds with a glottal chink

The paper presents extensions of the single-matrix formulation (Mokhtari et al., 2008, Speech Comm. 50(3) 179 – 190) that enable self-oscillation models of vocal folds, including glottal chink, to be connected to the vocal tract. They also integrate the case of a local division of the main air path into two lateral channels, as it may occur during the production of lateral consonants. Provided extensions are detailed by a reformulation of the acoustic conditions at the glottis, and at the upstream and downstream connections of bilateral channels. The simulation framework is validated through numerical simulations. The introduction of an antiresonance in the transfer function due to the presence of asymmetric bilateral channels is confirmed by the simulations. The frequency of the antiresonance agrees with the theoretical predictions. Simulations of static vowels reveal that the behavior of the vocal folds is qualitatively similar whether they are connected to the single-matrix formulation or to the classic reflection-type line analog model. Finally, the acoustic effect of the glottal chink on the production of vowels is highlighted by the simulations: the shortening of the vibrating part of the vocal folds lowers the amplitude of the glottal flow, and therefore lowers the global acoustic level radiated at the lips. It also introduces an offset in the glottal flow waveform.

The Vocal Tract Organ and the Vox Humana organ stop

Abstract The Vocal Tract Organ is a new musical instrument that consists of three-dimensional (3D)-printed vocal tracts (throat and mouth) for individual vowels sitting on loudspeakers to enable static vowel sounds to be produced. The acoustic excitation from the loudspeakers is a synthesized version of the typical waveform produced by the vibrating human vocal folds during pitched sounds, which enables the instrument to be played from a keyboard. The synthesis engine is implemented in Pure Data. This article describes the implementation of the Vocal Tract Organ, how it has been used musically to date, and the similarities and differences between it and the Vox Humana pipe organ stop. It is suggested that the Vocal Tract Organ will become an instrument in its own right and that it could be used as a direct replacement for the Vox Humana organ stop, given that its acoustic output is a much closer representation of the human vocal output than that from a Vox Humana organ pipe. In addition, the author makes use of the 3D-printed tracts in vocal and choral workshops as well as degree-level music technology education.

Temporal characteristics of vowel articulation: Electromyographic investigation of the “articulatory period”

In speech, articulator movement precedes the acoustic signal [e.g., Meyer (1991); Gracco (1988); Lubker and Gay (1982); Bell-Berti and Harris (1981)]. The onset and offset of movement, i.e., the ‘articulatory period' (Bell-Berti and Harris, 1981, p. 13) is said to encompass the acoustic signal however the precise timing and variability of this period are not well understood. Using electromyography (EMG), we examine the onset/offset of the articulatory period as a function of phonetic context. We recorded EMG activity from the posterior and anterior regions of the genioglossus (GG) muscle of the tongue in subjects during the articulation of static vowels, static vowels with initial coronal or palatal fricatives and vowels embedded in the nonce word [∫pVp]. We show front and high vowels entail earlier onset of GG activation and that vowels preceded by coronal fricatives are associated with significantly earlier muscle activation than vowels preceded by palatal fricatives. We also note timing differences between GG regions. Thus, for these stimuli, posterior GG EMG activation encompasses the duration of anterior GG EMG activation. These findings underscore the dynamic nature of lingual movement wherein regional tongue muscle activities exhibit predictable differences determined in large part by phonetic context.

Speech energy redistribution for intelligibility improvement in noise based on a perceptual distortion measure

A speech pre-processing algorithm is presented that improves the speech intelligibility in noise for the near-end listener. The algorithm improves intelligibility by optimally redistributing the speech energy over time and frequency according to a perceptual distortion measure, which is based on a spectro-temporal auditory model. Since this auditory model takes into account short-time information, transients will receive more amplification than stationary vowels, which has been shown to be beneficial for intelligibility of speech in noise. The proposed method is compared to unprocessed speech and two reference methods using an intelligibility listening test. Results show that the proposed method leads to significant intelligibility gains while still preserving quality. Although one of the methods used as a reference obtained higher intelligibility gains, this happened at the cost of decreased quality. Matlab code is provided.

A New Technique for Perceptual Distortion Measure on A Spectro Temporal Auditory Model

Speech processing is used widely in every day’s applications that most people take for granted, such as network wire lines, cellular telephony, telephony system and telephone answering machines. Due to its popularity and increasing of demand, engineers are trying various approaches of improving the process. One of the methods for improving is trying on different methods of filtering techniques. Thus, this instigates an introduction of a filtering technique known as Kalman filtering. In the early days, Kalman filtering was very popular in the research field of navigation because of its magnificent accurate estimation characteristic. Since then, electronic engineers manipulate its advantages to useful purpose in speech processing. Consequently, today it had become a popular filtering technique for estimating and resolving redundant errors containing in speech. A speech pre-processing algorithm is presented to improve the speech intelligibility in noise for the near-end listener. The algorithm improves the intelligibility by optimally redistributing the speech energy over time and frequency for a perceptual distortion measure, which is based on a spectro-temporal auditory model. In contrast o spectral-only models, short-time information is taken into account. As a consequence, the algorithm is more sensitive to transient regions, which will therefore receive more amplification compared to stationary vowels. It is known from literature that changing the vowel-transient energy ratio is beneficial for improving speech intelligibility in noise. Objective intelligibility prediction results show that the proposed method has higher speech intelligibility in noise compared to other reference methods, without modifying the global speech energy.

Auditory Spectral Integration in the Perception of Static Vowels

To evaluate potential contributions of broadband spectral integration in the perception of static vowels. Specifically, can the auditory system infer formant frequency information from changes in the intensity weighting across harmonics when the formant itself is missing? Does this type of integration produce the same results in the lower (first formant [F1]) and higher (second formant [F2]) regions? Does the spacing between the spectral components affect a listener's ability to integrate the acoustic cues? Twenty young listeners with normal hearing identified synthesized vowel-like stimuli created for adjustments in the F1 region (/Λ/-/α/, /i/-/ε/) and in the F2 region (/Λ/-/æ/). There were 2 types of stimuli: (a) 2-formant tokens and (b) tokens in which 1 formant was removed and 2 pairs of sine waves were inserted below and above the missing formant; the intensities of these harmonics were modified to cause variations in their spectral center of gravity (COG). The COG effects were tested over a wide range of frequencies. Obtained patterns were consistent with calculated changes to the spectral COG, in both the F1 and F2 regions. The spacing of the sine waves did not affect listeners' responses. The auditory system may perform broadband integration as a type of auditory wideband spectral analysis.

High-Pitch Formant Estimation by Exploiting Temporal Change of Pitch

This paper considers the problem of obtaining an accurate spectral representation of speech formant structure when the voicing source exhibits a high fundamental frequency. Our work is inspired by auditory perception and physiological studies implicating the use of pitch dynamics in speech by humans. We develop and assess signal processing schemes aimed at exploiting temporal change of pitch to address the high-pitch formant frequency estimation problem. Specifically, we propose a 2-D analysis framework using 2-D transformations of the time-frequency space. In one approach, we project changing spectral harmonics over time to a 1-D function of frequency. In a second approach, we draw upon previous work of Quatieri and Ezzat , , with similarities to the auditory modeling efforts of Chi , where localized 2-D Fourier transforms of the time-frequency space provide improved source-filter separation when pitch is changing. Our methods show quantitative improvements for synthesized vowels with stationary formant structure in comparison to traditional and homomorphic linear prediction. We also demonstrate the feasibility of applying our methods on stationary vowel regions of natural speech spoken by high-pitch females of the TIMIT corpus. Finally, we show improvements afforded by the proposed analysis framework in formant tracking on examples of stationary and time-varying formant structure.

Vowel recognition from articulatory position time‐series data.

The goal of this project is to determine the accuracy and processing speed of different approaches for mapping time‐varying articulatory positional data to vowels. Three widely used classifiers were compared on two datasets: one single speaker and one multiple speaker. The single‐speaker dataset was acquired using the Articulograph AG500. The multiple‐speaker dataset was obtained from seven speakers in the Xray Microbeam Speech Production Database (Westbury, 1994). The recognition rate for single speaker dataset (eight English vowels) ranged from 94.25% to 98.1%, and from 62.38% to 99.35% for the multiple‐speaker dataset. For the single‐speaker dataset, recognition accuracy was comparable across classifiers. For the multiple‐speaker dataset, recognition accuracy was better for the Support Vector Machine and C4.5 than for the neural networks. The decision tree generated by C4.5 was consistent with articulatory features commonly used to descriptively distinguish vowels. Moreover, the Support Vector Machine and C4.5 performed much faster than did the neural network. The high recognition rates observed suggest that static vowels can be accurately recognized from articulatory position time‐series data. The findings are intended to improve the accuracy and response time of a real‐time articulatory‐movement based speech synthesizer.

Vowel Space Characteristics and Vowel Identification Accuracy

To examine the relation between vowel production characteristics and intelligibility. Acoustic characteristics of 10 vowels produced by 45 men and 48 women from the J. M. Hillenbrand, L. A. Getty, M. J. Clark, and K. Wheeler (1995) study were examined and compared with identification accuracy. Global (mean f0, F1, and F2; duration; and amount of formant movement) and fine-grained measures (vowel space area; mean distance among vowels; f0, F1, and F2 ranges; duration ratio between long and short vowels; and formant movement ratio between dynamic and static vowels) were used to predict identification scores. Acoustic measures of the most frequently confused pairs (/ae/-/epsilon/ and /a/-/inverted v/) were compared. Global and fine-grained measures accounted for less than 1/4 of variance in identification scores: Vowel space area alone accounted for 9%-12% of variance. Differences in vowel identification were largely due to poor identification of /ae/, /epsilon/, /a/, or /inverted v/. Well-identified vowels were distinctive in formant frequencies, duration, and amount of formant movement over time. Distinctiveness among neighboring vowels is more important in determining vowel intelligibility than vowel space area. Acoustic comparison of confused vowels may be more useful in studying intelligibility of normal and disordered speech than in measuring vowel space area.

Acquired Dysarthria within the Context of the Four-level Framework of Speech Sensorimotor Control

The Four-Level Framework of speech sensorimotor control (Van der Merwe, 1997) complicates the traditional view of dysarthria as a purely motor execution disorder. According to this framework, hypokinetic, hyperkinetic and ataxic dysarthria are programming-execution dysarthrias, while flaccid dysarthria is the only execution dysarthria. This preliminary study aimed to differentiate programming-execution dysarthria from execution dysarthria by examining variability of the temporal control of speech. Six participants and five control participants repeated 15 stimulus words ten times. Voice onset time, vowel duration, vowel steady state duration and vowel formant transition duration were measured acoustically. The coefficient of variation of the temporal parameters, and the correlation coefficient between the durational parameters, were calculated and analysed using descriptive statistics. The coefficient of variation revealed that the speakers with dysarthria were more variable than the control speakers. All participants, except those with flaccid dysarthria, showed similar patterns of intra-subject variability. Those with flaccid dysarthria exhibited greater intra-subject variability of voice onset time. The correlation analysis did not reveal differences between dysarthria type, or between the dysarthric speakers and the controls. Differences found in the patterns of variability may support the hypothesis that individuals with programming-execution dysarthria resort to a different level of control than those with execution dysarthria. Further research in this field is necessary.

Speech intelligibility and the vowel space: Normal speech

Vowel formant frequency values and related vowel space measures have been widely used in the study of speech intelligibility for normal and disordered talkers. The strength of vowel space measures in predicting speech intelligibility for normal talkers, however, is relatively low: [Bradlow et al. (1996)] found that vowel space dispersion predicted 19% of variance in sentence intelligibility, and [Hazan and Markham (2004)] found that the F2 difference between /i/ and /u/ predicted 16% of variance in word intelligibility scores. Using data from the 45 men and 48 women speakers in the Hillenbrand et al. (1995) database, acoustic characteristics of ten vowels were used to predict identification accuracy. Multiple regressions revealed that global measures (mean f0, F1 and F2, duration, and amount of formant movement) and fine-grained measures (vowel space area, mean distance among vowels, ranges for f0, F1, and F2, duration ratio between long and short vowels, and dynamic ratio between dynamic and static vowels) accounted for less than one-quarter of the variance in identification scores across talkers. Focusing on confusions among spectrally similar vowels may provide better information on intelligibility differences among normal talkers. Goodness ratings may provide a wider range of scores for statistical analysis than identification accuracy.

Time dependence of vocal tract modes during production of vowels and vowel sequences

Vocal tract shaping patterns based on articulatory fleshpoint data from four speakers in the University of Wisconsin x-ray microbeam (XRMB) database [J. Westbury, UW-Madison, (1994)] were determined with a principal component analysis (PCA). Midsagittal cross-distance functions representative of approximately the front 6 cm of the oral cavity for each of 11 vowels and vowel-vowel (VV) sequences were obtained from the pellet positions and the hard palate profile for the four speakers. A PCA was independently performed on each speaker's set of cross-distance functions representing static vowels only, and again with time-dependent cross-distance functions representing vowels and VV sequences. In all cases, results indicated that the first two orthogonal components (referred to as modes) accounted for more than 97% of the variance in each speaker's set of cross-distance functions. In addition, the shape of each mode was shown to be similar across the speakers suggesting that the modes represent common patterns of vocal tract deformation. Plots of the resulting time-dependent coefficient records showed that the four speakers activated each mode similarly during production of the vowel sequences. Finally, a procedure was described for using the time-dependent mode coefficients obtained from the XRMB data as input for an area function model of the vocal tract.

Vowel identification and vowel space characteristics

Acoustic characteristics of ten vowels produced by 45 men and 48 women from the Hillenbrand et al. (1995) study were correlated with identification accuracy. Global (mean f0, F1 and F2, duration, and amount of formant movement) and distinctive measures (vowel space area, mean distance among vowels, f0, F1 and F2 ranges, duration ratio between long and short vowels, and dynamic ratio between dynamic and static vowels) were used to predict identification scores. Global and distinctive measures accounted for less than one-fourth of variance in identification scores: vowel space area alone accounted for 9% to 12% of variance. Differences in vowel identification across talkers were largely due to poor identification of two spectrally similar vowel pairs /ae/-/eh/ and /uh/-/ah/. Results of acoustic analysis and goodness ratings for well-identified and poorly identified versions of these vowels will be presented. Preliminary analysis revealed that well-identified vowels differed from poorly identified tokens not only in static formant frequencies but also in duration and amount of formant movement over time. A talker’s ability to create distinctions among neighboring vowels is more important in determining vowel intelligibility than overall measures such as vowel space area.

Real-Time Dynamic Articulations in the 2-D Waveguide Mesh Vocal Tract Model

Time domain articulatory vocal tract modeling in one-dimensional (1-D) is well established. Previous studies into two-dimensional (2-D) simulation of wave propagation in the vocal tract have shown it to present accurate static vowel synthesis. However, little has been done to demonstrate how such a model might accommodate the dynamic tract shape changes necessary in modeling speech. Two methods of applying the area function to the 2-D digital waveguide mesh vocal tract model are presented here. First, a method based on mapping the cross-sectional area onto the number of waveguides across the mesh, termed a widthwise mapping approach is detailed. Discontinuity problems associated with the dynamic manipulation of the model are highlighted. Second, a new method is examined that uses a static-shaped rectangular mesh with the area function translated into an impedance map which is then applied to each waveguide. Two approaches for constructing such a map are demonstrated; one using a linear impedance increase to model a constriction to the tract and another using a raised cosine function. Recommendations are made towards the use of the cosine method as it allows for a wider central propagational channel. It is also shown that this impedance mapping approach allows for stable dynamic shape changes and also permits a reduction in sampling frequency leading to real-time interaction with the model

Salience of virtual formants as a function of the frequency separation between spectral components

The center-of-gravity (COG) hypothesis proposed by Chistovich and others for the perception of static vowels suggests that auditory spectral integration may occur when two or more formants fall within a 3.5 bark bandwidth. While several studies have examined the bandwidth limits of such integration, this study examines the extent to which spectral integration is uniform within this putative 3.5-bark range. We examine the perceptual salience of virtual formants produced by modifying the spectral COG of two closely spaced narrow-bandwidth resonances. Three different vowel series were created: [i-■], [■-■] and [■-■]. A second set of vowels was then created in which one of the formants (F1 in [i-■], F2 in [■-■] and F3 in [■-■]) was replaced by a virtual formant whose COG matched that of the formant that had been removed. The frequency separation between the two component resonances was then systematically varied between 1.5 and 3.5 barks and a singleinterval 2AFC vowel identification task was used to obtain estimates of vowel quality for each series step. Results will be discussed in terms of whether the spectral integration effects within the 3.5 bark decline as the frequency separation between the resonance components increases. [Work supported by NIDCD R01DC00679-01A1.]