Vocal Tract Model Research Articles

An acoustic glottal source for vocal tract physical models

A sound source is proposed for the acoustic measurement of physical models of the human vocal tract. The physical models are produced by fast prototyping, based on magnetic resonance imaging during prolonged vowel production. The sound source, accompanied by custom signal processing algorithms, is used for two kinds of measurements from physical models of the vocal tract: (i) amplitude frequency response and resonant frequency measurements, and (ii) signal reconstructions at the source output according to a target pressure waveform with measurements at the mouth position. The proposed source and the software are validated by computational acoustics experiments and measurements on a physical model of the vocal tract corresponding to the vowels [] of a male speaker.

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.

Radius vector-driven 3-D Mandarin vocal tract model

Radius vector-driven 3-D Mandarin vocal tract model

Relevance of the Implementation of Teeth in Three-Dimensional Vocal Tract Models.

Recently, efforts have been made to investigate the vocal tract using magnetic resonance imaging (MRI). Due to technical limitations, teeth were omitted in many previous studies on vocal tract acoustics. However, the knowledge of how teeth influence vocal tract acoustics might be important in order to estimate the necessity of implementing teeth in vocal tract models. The aim of this study was therefore to estimate the effect of teeth on vocal tract acoustics. The acoustic properties of 18 solid (3-dimensional printed) vocal tract models without teeth were compared to the same 18 models including teeth in terms of resonance frequencies (fRn). The fRn were obtained from the transfer functions of these models excited by white noise at the glottis level. The models were derived from MRI data of 2 trained singers performing 3 different vowel conditions (/i/, /a/, and /u/) in speech and low-pitched and high-pitched singing. Depending on the oral configuration, models exhibiting side cavities or side branches were characterized by major changes in the transfer function when teeth were implemented via the introduction of pole-zero pairs. To avoid errors in modeling, teeth should be included in 3-dimensional vocal tract models for acoustic evaluation. https://doi.org/10.23641/asha.5386771.

A new metric for calculating acoustic dispersion in stop inventories

Dispersion Theory (DT; Liljencrants and Lindblom, 1972) claims that acoustically dispersed vowel inventories should be typologically common. Literature on DT has focused on vowels, where the predictions are robust, and less work has been done on consonants. This paper uses vocal tract model data of stops (as in Schwartz et al. 2012) to extend the predictions of DT to consonants, revealing problems with the conventional method of calculating dispersion. Dispersion is often quantified using triangle area between three category means as points in acoustic space. This approach ignores distributions and reduces entire acoustic categories (which have large variances and different distribution shapes) to single points. Within-category variance is a factor in DT (Lindblom, 1986) and experimental data shows that it affects perception (Clayards 2008), yet conventional dispersion metrics do not take it into account. Here, a new metric based on the Jeffries-Matusita distance is proposed and compared with the more conventional mean to mean distance approach. The incorporation of covariance better reflects human perception, which has implications for considering dispersion in any acoustic space. Nevertheless, this does not recover the predictions of DT, suggesting DT does not apply to consonants and vowels in the same way.

Modeling individual vocal differences in group-living lemurs using vocal tract morphology.

Vocal individuality is widespread in social animals. Individual variation in vocalizations is a prerequisite for discriminating among conspecifics and may have facilitated the evolution of large complex societies. Ring-tailed lemurs Lemur catta live in relatively large social groups, have conspicuous vocal repertoires, and their species-specific utterances can be interpreted in light of source-filter theory of vocal production. Indeed, their utterances allow individual discrimination and even recognition thanks to the resonance frequencies of the vocal tract. The purpose of this study is to determine which distinctive vocal features can be derived from the morphology of the upper vocal tract. To accomplish this, we built computational models derived from anatomical measurements collected on lemur cadavers and compared the results with the spectrographic output of vocalizations recorded from ex situ live individuals. Our results demonstrate that the morphological variation of the ring-tailed lemur vocal tract explains individual distinctiveness of their species-specific utterances. We also provide further evidence that vocal tract modeling is a powerful tool for studying the vocal output of non-human primates.

Cantor Digitalis: chironomic parametric synthesis of singing

Cantor Digitalis is a performative singing synthesizer that is composed of two main parts: a chironomic control interface and a parametric voice synthesizer. The control interface is based on a pen/touch graphic tablet equipped with a template representing vocalic and melodic spaces. Hand and pen positions, pen pressure, and a graphical user interface are assigned to specific vocal controls. This interface allows for real-time accurate control over high-level singing synthesis parameters. The sound generation system is based on a parametric synthesizer that features a spectral voice source model, a vocal tract model consisting of parallel filters for vocalic formants and cascaded with anti-resonance for the spectral effect of hypo-pharynx cavities, and rules for parameter settings and source/filter dependencies between fundamental frequency, vocal effort, and formants. Because Cantor Digitalis is a parametric system, every aspect of voice quality can be controlled (e.g., vocal tract size, aperiodicities in the voice source, vowels, and so forth). It offers several presets for different voice types. Cantor Digitalis has been played on stage in several public concerts, and it has also been proven to be useful as a tool for voice pedagogy. The aim of this article is to provide a comprehensive technical overview of Cantor Digitalis.

口腔単純形状モデルを用いた摩擦音/s/と/sh/の空力音響解析

口腔単純形状モデルを用いた摩擦音/s/と/sh/の空力音響解析

An acoustically-driven vocal tract model for stop consonant production

The purpose of this study was to further develop a multi-tier model of the vocal tract area function in which the modulations of shape to produce speech are generated by the product of a vowel substrate and a consonant superposition function. The new approach consists of specifying input parameters for a target consonant as a set of directional changes in the resonance frequencies of the vowel substrate. Using calculations of acoustic sensitivity functions, these “resonance deflection patterns” are transformed into time-varying deformations of the vocal tract shape without any direct specification of location or extent of the consonant constriction along the vocal tract. The configuration of the constrictions and expansions that are generated by this process were shown to be physiologically-realistic and produce speech sounds that are easily identifiable as the target consonants. This model is a useful enhancement for area function-based synthesis and can serve as a tool for understanding how the vocal tract is shaped by a talker during speech production.

A Modeling Study of the Effects of Vocal Tract Movement Duration and Magnitude on the F2 Trajectory in CV Words

This study used a computational vocal tract model to investigate the relationship of diphthong duration and vocal tract movement magnitude to measures of the F2 trajectory in CV words. Three words (bough, boy, and buy) were simulated on the basis of an adult female vocal tract model, in which the model parameters were estimated from audio recordings of a female talker. Model parameters were then modified to generate 35 simulations of each word corresponding to 7 different durations and 5 movement magnitude settings. In addition, these simulations were repeated with vocal tract lengths representative of an adult male and an approximately 6-year-old child. On the basis of univariate analysis, measures of frequency predicted changes in magnitude, and temporal measures predicted changes in speaking rate consistent with the hypothesis. The combined effects of duration and magnitude showed that F2 was more sensitive to changes in magnitude at shorter word durations compared with longer word durations. This finding held across words and vocal tract length. Results suggest that there is an interaction between duration and magnitude that affects the slope of the F2 trajectory. The next step is to relate kinematics to F2 trajectory output using real speakers.

Influence of the Human Body Mass in the Open-air MRI on Acoustic Noise Spectrum

The paper analyses changes in spectral properties of the acoustic noise when the examined person lies in the scanning area of the open-air magnetic resonance imager (MRI) and consequently the holder of the lower gradient coils is loaded with the mechanical mass represented by the person’s weight. The acoustic noise pressure level is mapped in the MRI neighborhood, too. Obtained results of spectral analysis will be used for design of a correction filter to suppress the noise in the simultaneously recorded speech signal for 3D modelling of the human vocal tract.

Identification of stop consonants produced by an acoustically-driven model of a child-like vocal tract

A model of a child-like vocal tract has been developed such that the deformation patterns superimposed on a vowel substrate to generate coarticulated consonants are specified by a time-varying set of directional shifts in the first three resonance frequencies. These deflection patterns are denoted as a combination of three numbers each of which can vary between -1 and 1; a negative value implies a downward shift in a resonance frequency whereas an upward shift results for positive value. For example, a “bilabial” consonant specified as [-1,-1,-1] would be transformed via calculations of acoustic sensitivity functions to a time-varying vocal tract shape that presents the expected constriction at the lips, but also modifies other parts of the vocal tract that may be necessary for producing the appropriate formant transitions into and out of the consonant. Using this model, three sets of 30 VCV utterances were generated in which the values of deflection patterns were set to produce vocal tract shapes that hy...

The UCLA voice synthesizer, version 2.0

Based on our psychoacoustic model of voice quality, the UCLA voice synthesizer allows users to copy synthesize nearly any steady-state voice sample or to create stimuli that systematically vary in specific acoustic dimensions. This new release contains a number of significant improvements from earlier versions. The vocal tract model now includes three spectral zeros with adjustable bandwidths. The precise spectral shape of the harmonic and inharmonic sources can be modified at will, either by adjusting the heights of the harmonics or the noise amplitude in a selected frequency range or by specifying the desired spectral or noise slope in a range. Any number of ranges of any size can be specified. Synthesizer variables can be individually saved and copied between cases. Additional changes increase control of analysis parameters and the ease with which series of stimuli can be created. The synthesizer is fully documented and is freely available for download from headandnecksurgery.ucla.edu/glottalaffairs. Copies will also be available at the conference. [Research supported by NIH.]

Goodness evaluation by L1 Spanish speakers of various allophones, made by human and by vocal tract models, of the Spanish phonemes /l/ and /r/

The present study reports the results of two perceptive experiments, in which L1 speakers of Peninsular Spanish were asked to listen to a series of sound stimuli and to evaluate the goodness of each stimulus as a realization of Spanish consonant /l/ or /r/. The stimuli were made of natural human voice in the first experiment, and with a set of vocal tract models in the second experiment. The results were similar for the two experiments. The lateral approximant stimuli were highly evaluated as allophones of /l/. Some lateral approximant stimuli were longer than others, but the duration difference caused little effect on the evaluation. The flap stimuli obtained low scores as /l/. Conversely, for the phoneme /r/, only the flap stimuli acquired high evaluations, while all the lateral approximant stimuli obtained extremely low scores. The retroflex approximant stimuli, similar to English /r/, obtained low evaluations for both /l/ and /r/ in the first experiment using a human voice, while they were evaluated slightly higher as /l/ than as /r/ in the second experiment using the vocal tract models. These results are expected to provide L2 Spanish learners with useful knowledge in their pronunciation learning.

Tools for education in acoustics and phonetics

We have developed a series of tools for education in acoustics and phonetics. Some of them are in electronic form, while some are physical models. A set of demonstrations and explanations in acoustic phonetics is now available as “Acoustic-Phonetics Demonstrations (APD)” through our WebSite: www.splab.net/APD/. We are adding content and have made YouTube video clips using physical models available on our WebPages. In particular, we use the physical models of the human vocal tract to demonstrate acoustic phenomena and theories. They are useful tools for explaining acoustics in speech production, such as the source-filter theory and the relationship between vocal-tract configurations and vowel quality. Furthermore, the models are effective when teaching articulation/pronunciation of native and non-native languages. Due to the high demand for vocal-tract models of various sounds in multiple languages, we are expanding the range of sounds to include more than Japanese sounds, but also additional vowels and consonants in different languages, including English. We also discuss how to apply three-dimensional printing.

Study on the principle of sound resonance in human whistling using physical models of a human vocal tract

The principle behind sound production in human whistling is relatively unknown. A good understanding of this principle would be beneficial to both trainer and trainee. Rayleigh identified whistling frequency as being determined by the mouth cavity and pointed out that earlier ideas that relate the sound-producing mechanism to the vibration of the lips are inaccurate. Wilson et al. reported that the human whistling resonant frequency is close to the Helmholtz resonant frequency, and through some physical-model-based experiments, they determined that the resonator can be excited by a flow through the smooth-edged orifices bounding the resonant cavity. Thus, the principles of whistling in terms of articulation have been reported to be based on the Helmholtz resonance. However, some whistlers can produce high-pitched sounds by blowing harder without changing the capacity of the resonance chamber, which is similar to a high-pitched sound produced by a wind instrument by air-column resonance. This work studies ...

Effect of tongue position in the simplified vocal tract model of sibilant fricatives /s/ and /ʃ/

In this study, a simplified vocal tract model which reproduces both sibilant fricatives /s/ and /ʃ/ is proposed to investigate the effect of tongue position on flow and sound generation in the vocal tract. The simplified model consists of a tongue with sibilant groove and upper and lower teeth in a rectangular channel. Dimensions of the model were determined based on CT images of the subject pronouncing sibilant /s/. The sound and flow fluctuation generated by the model were measured with a microphone and a hot-wire anemometer, respectively. In addition, the large-eddy simulation was conducted to reveal the characteristics of sound source generated by the turbulent flow in the vocal tract. The acoustic measurement showed that the backward shift of the tongue increased the spectral slope value, indicating that the acoustic characteristics were changed from /s/ to /ʃ/. The large-eddy simulation showed that the position of sound source downstream of teeth was shifted backward and the magnitude of the sound source was decreased by shifting the tongue backward. These results indicate that both the resonance characteristics and the position of the sound source were changed by shifting the tongue which causes the difference in acoustic characteristics between /s/ and /ʃ/.

Speech biomechanics: What have we learned and modeled since Joseph Perkell’s tongue model In 1974?

With his “physiologically oriented, dynamic model of the tongue, Joseph Perkell introduced in 1974 a new methodological approach to understanding the “relationships among phonetic models and the properties and capabilities of the speech-production mechanism.” This approach has guided a large part of our studies in the two last decades. In order to investigate how mechanical properties of the orofacial motor system constrain the degrees of freedom of speech articulation and contribute to shaping the speech signals exchanged between speakers and listeners, we, among other research groups, have developed increasingly more realistic 2D, and then 3D, finite element(FE) biomechanical models of the human vocal tract and face. After summarizing some of our modeling and simulation results that shed light on some basic characteristics of speech production, we present recent developments which aim to improve the realism of the models: evaluation of the links between the FE mesh structure (based either on tetrahedra, hexahedra, or mixed elements) and simulation accuracy; development of an active 3D element that simulates muscle mechanics and muscle force generation mechanisms; use of Diffusion Tensor Imaging to investigate muscle anatomy; design of an Atlas-based method (i.e., without manual image segmentation) for the automatic generation of subject-specific models.

Vocal-tract models and their applications in education for intuitive understanding of speech production

Vocal-tract models and their applications in education for intuitive understanding of speech production

Principles of sound resonance in human whistling using physical models of human vocal tract

Principles of sound resonance in human whistling using physical models of human vocal tract