Vocal Tract Model Research Articles

Evaluation of Upper Airway in Obstructive Sleep Apnea: Synchronized Imaging and Acoustic Analyses

Objectives: Snore signals are produced by soft tissue vibration in the upper airway (UA). To date, there is no gold standard to localize UA narrowing sites in patients with obstructive sleep apnea syndrome (OSAS). In this study, we use a “vocal tract model” to investigate the association between snore acoustic characteristics and UA collapse. Dynamic magnetic resonance imaging (dMRI) was used to provide UA anatomy information. Methods: We report data from 44 polysomnography (PSG)-confirmed snoring episodes. Dynamics of UA and acoustic manifestations of snore events during natural sleep were recorded with synchronized dMRI and fiber-optic microphone recording. After the deletion of background dMRI noises, the spectrum envelope was reconstructed by applying a vocal tract model to characterize the snoring patterns. The temporal change of spectrum envelope was analyzed using Gabor transformation to correlate with UA dynamics. Results: The snore events were categorized into 28 none, 7 typical (velopharyngeal), and 9 mixed (velopharyngeal and hypopharyngeal) obstructions. The snore’s dominant frequency of envelop spectrum can differentiate none from typical UA obstructions (82.3±16.4Hz vs 101.9±9.1 Hz, p<0.01); and can differentiate mixed from typical obstructions (131.9±19.8 Hz vs.101.9±9.1 Hz, p < 0.05) Conclusions: The dynamics of acoustic patterns derived from the envelope of Gabor spectrogram are unique and characteristics in different types of obstruction with the advantage of low-cost and easy availability. Therefore, the dynamics of acoustic patterns warrants further investigation to identify the structural changes of the obstruction sites in OSAS patients.

Applied aspects of the synthesis and analysis of voice information

The authors present new results in solving problems of concatenative segment synthesis of voice information with prosody and vocal utterance, computer modeling of human voice signals based on joint models of human voice source and vocal tract, and speech signal preprocessing for automated documenting systems. The experiments show the efficiency of the proposed approaches.

Mechanical bent-type models of the human vocal tract consisting of blocks

In our previous work, we developed several physical models of the human vocal tract and reported that they are intuitive and helpful for students studying acoustics and speech science. Models with a bent vocal tract can achieve relatively realistic tongue movements. These bent-type models had either a flexible tongue or a sliding tongue. In the former case, the tongue was made of a flexible gel-type material so that we could form arbitrary tongue shapes. However, this flexibility meant that training is needed to achieve target sounds. In the latter case, the tongue was made of an acrylic resin, and only a limited number of vowel sounds can be achieved because so few sliding parts are available to change the tongue shape. Therefore, in this study, we redesigned the mechanical bent-type models so that they now consist of blocks. By placing the blocks at the proper positions, the block-type model can produce five intelligible Japanese vowels. We also designed a single bent-type model with sliding blocks that can produce several vowel sounds. [This work was partially supported by a Grant-in-Aid for Scientific Research (24501063) from the Japan Society for the Promotion of Science.]

Improved vocal tract reconstruction and modeling using an image super-resolution technique

Magnetic resonance imaging has been widely used in speech production research. Often only one image stack (sagittal, axial, or coronal) is used for vocal tract modeling. As a result, complementary information from other available stacks is not utilized. To overcome this, a recently developed super-resolution technique was applied to integrate three orthogonal low-resolution stacks into one isotropic volume. The results on vowels show that the super-resolution volume produces better vocal tract visualization than any of the low-resolution stacks. Its derived area functions generally produce formant predictions closer to the ground truth, particularly for those formants sensitive to area perturbations at constrictions.

Three-dimensional vocal tract modeling of fricatives /s/ and /sh/ for post-glossectomy speakers

Production of fricatives involves a narrow supraglottal constriction along the vocal tract. Air flows through the constriction, and generates turbulent noise source(s) by impinging on some obstacles downstream. In post-glossectomy speakers, the production of /s/ and /sh/ is often problematic. It is mainly caused by the tongue surgery, which changes tongue properties such as volume, motility, and symmetry, preventing the tongue from creating proper constrictions. The purpose of this study was to gain some insights on how the vocal tracts of abnormal /s/ and /sh/ are shaped and what are their corresponding acoustic consequences. Based on cine magnetic resonance images, we built 3-D vocal tract models for /s/ and /sh/ from two post-glossectomy speakers (one with abnormal /s/ and the other with abnormal /sh/). Due to the missing part of the tongue, the reconstructed vocal tracts are asymmetric with either an air-flow bypass or a side branch formed near the constrictions. Two coupled physics submodels are included in the 3-D FEM acoustic simulation: incompressible potential flow for the mean air flow and aeroacoustics for the distributed noise sources. The resulting acoustic spectra and acoustic roles of air flow bypass or side branch will be discussed. [This study was supported by NIH R01CA133015.]

Learning acoustic phonetics by listening, seeing, and touching

There is a huge volume of written textbooks available in virtually every modern field, including acoustic phonetics. However, in areas dealing with acoustics, learners often face problems and limitations when they deal with only written material and no audio or visual information. As one response to this problem, we have developed several sets of physical models of the human vocal tract and have shown that they are extremely useful for intuitive understanding. In addition, we also developed a tool called “Digital Pattern Playback.” Another solution is an online version featuring demonstrations. We are currently collecting materials, mainly in the form of sounds, for educational purposes in acoustics and phonetics and are releasing them as “Acoustic-Phonetics Demonstrations” through our Web site. These demonstrations are designed for students in linguistics, phonetics and phonology, speech pathology, audiology, psychoacoustics, speech engineering, and others. However, potential users are not limited to these groups, as we feel that a wide range of learners can obtain tremendous benefits from the demonstrations, including those who are studying foreign languages or patients undergoing speech articulation therapy. [Work partially supported by a Grant-in-Aid for Scientific Research (24501063) from the Japan Society for the Promotion of Science.]

Perceptual consequences of changes in epilaryngeal area and shape

Decreasing epilaryngeal area has been shown to increase glottal flow pulse skewing and harmonic amplitudes [Titze, J. Acoust. Soc. Am. 123, 2733 (2008)]. It is not known, however, whether listeners perceive voice quality changes when epilaryngeal area is altered, or if perceived quality is different if the area change occurs at the ventricular folds or aryepiglottic (AE) folds. In this study, a kinematic vocal tract model was used to create five epilaryngeal cavity shapes resulting from constriction and retraction of the ventricular and AE folds. Four voice sources simulating varying degrees of vocal deviation were filtered through the five shapes for a total of 20 stimuli. Fourteen listeners completed a sort and rate task. Results were analyzed using multidimensional scaling (MDS). Altering the epilaryngeal cavity shape resulted in voice quality differences, and perceptual distances differed by voice source. AE fold constriction was perceived most differently from other shapes for all talkers. Ventricular fold constriction was perceived most similar to AE constriction for 3 of the 4 voice sources. Glottal flow and acoustic differences for each epilaryngeal shape will be described and related to the perceived differences in voice quality.

Nonlinear Source-Filter Coupling Due to the Addition of a Simplified Vocal Tract Model for Excised Larynx Experiments

Traditional excised larynx dissection and setup calls for the removal of all supraglottal structures, eliminating any source-filter interactions that measurably affect the acoustic properties of phonation. We introduce a simplified vocal tract model that can be used in the excised larynx experiments and tested the nonlinear source-filter interactions that are present with the addition of highly coupled, supraglottal structures. Aerodynamic and acoustic data were measured at phonation threshold pressure (PTP) and +25% PTP in 10 excised canine larynges using a modified dissection technique. PTP and phonation threshold flow (PTF) were defined as the pressure and flow at the phonation onset; phonation threshold power (PTW) is the product of these values. Data were recorded for four experimental conditions: PTP without vocal tract; +25% PTP without vocal tract; PTP with vocal tract; and +25% PTP with vocal tract. Differences in PTP, PTF, and PTW were evaluated. For trials conducted at +25% PTP, differences in airflow were evaluated. PTP (P=0.009) and PTW (P=0.002) were significantly reduced with the addition of the novel vocal tract. A reduction in PTF was also present with the vocal tract (P=0.021), but airflow was not significantly reduced in +25% PTP trials (P=0.196). The proposed vocal tract can be used with complete larynges when conducting excised larynx experiments. The effects of nonlinear source-filter interaction were observed during trials with the vocal tract, as evidenced by changes in threshold aerodynamic parameters.

Investigation of the Effect of Obstacle Placed Near the Human Glottis on the Resultant Sound Pressure

Simulation of human vocal tract model is necessary to understand the effect of different conditions on speech generation. In natural calamities such as earthquakes person may swallow soil due to large amount of dust around. The effect of soil is investigated on the sound pressure in the human vocal tract. The generation of speech starts from the glottis, so the soil obstacle is positioned near it. The whole model is designed and analyzed using FEM. As the speech signal requires a bandwidth near about 4 kHz, the investigation carried out to obtained sound pressure in the human vocal tract from 100 -5000 Hz sound signals.

Three-Dimensional Digital Waveguide Mesh Simulation of Cylindrical Vocal Tract Analogs

3D time-domain acoustic modeling techniques have the potential to produce more accurate simulation of the vocal tract than previously implemented 1D or 2D solutions, although the variability of human voice renders it a problematic benchmark for validation of its resynthesis. This study uses acoustic measurement of acrylic cylindrical vocal tract models derived from X-Ray data to assess the validity of comparable 3D digital waveguide mesh simulations. It is found that for more simple structures the 3D digital waveguide mesh is able to reproduce the acoustic behavior up to 10 kHz with only slight errors in resonant frequencies. As the simulated structures become more geometrically complex, this shifting becomes more severe.

An ANN based approach to recognize initial phonemes of spoken words of Assamese language

Initial phoneme is used in spoken word recognition models. These are used to activate words starting with that phoneme in spoken word recognition models. Such investigations are critical for classification of initial phoneme into a phonetic group. A work is described in this paper using an artificial neural network (ANN) based approach to recognize initial consonant phonemes of Assamese words. A self organizing map (SOM) based algorithm is developed to segment the initial phonemes from its word counterpart. Using a combination of three types of ANN structures, namely recurrent neural network (RNN), SOM and probabilistic neural network (PNN), the proposed algorithm proves its superiority over the conventional discrete wavelet transform (DWT) based phoneme segmentation. The algorithm is exclusively designed on the basis of Assamese phonemical structure which consists of certain unique features and are grouped into six distinct phoneme families. Before applying the segmentation approach using SOM, an RNN is used to take some localized decision to classify the words into six phoneme families. Next the SOM segmented phonemes are classified into individual phonemes. A two-class PNN classification is performed with clean Assamese phonemes, to recognize the segmented phonemes. The validation of recognized phonemes is checked by matching the first formant frequency of the phoneme. Formant frequency of Assamese phonemes, estimated using the pole or formant location determination from the linear prediction model of vocal tract, is used effectively as a priori knowledge in the proposed algorithm.

Reconsideration on the Principle of Sound Production in Human Whistling using the Vocal Tract Model

The principle of sound production in human whistling is not well-known. The purpose of this letter is reconsideration on the principle of sound production in human whistling using physical models of the human vocal tract based on the X-ray CT data. This letter shows that the whistling resonant frequency is very near the formant resonance frequency.

Mechanical Models of the Human Vocal Tract

Mechanical Models of the Human Vocal Tract

A parametric method of computing acoustic characteristics of simplified three-dimensional vocal-tract model with wall impedance

A parametric method of computing acoustic characteristics of simplified three-dimensional vocal-tract model with wall impedance

Vocal tract modeling techniques: from human voice to non-human primates vocalizations

Vocal tract modeling techniques: from human voice to non-human primates vocalizations

Vocal tract modeling techniques: from human voice to non-human primates vocalizations

Vocal tract modeling techniques: from human voice to non-human primates vocalizations

Improving vocal tract reconstruction and modeling through super-resolution volume reconstruction technique

Magnetic resonance imaging has been widely used in speech production for vocal tract reconstruction and modeling. In order to observe detailed structures in the vocal tract, three orthogonal image stacks (sagittal, coronal, and axial) are usually acquired. Due to many constraints, each stack typically has an in-plane resolution which is much better than the out-of-plane resolution. Usually vocal tract modeling is based on just one of these three stacks. As a result, additional useful information revealed by the other two datasets is excluded in the vocal tract model. This study is to improve the vocal tract reconstruction and modeling by integrating information from all of the three stacks. To do so, a super-resolution reconstruction method recently developed to generate an isotropic image volume is used to integrate the three orthogonal stacks. Based on the ATR MRI database of vowel production, vocal tract models from MR images in high resolution, low resolution (simulated through downsampling), and super-resolution were built respectively and compared. The improvement in vocal tract modeling due to the super-resolution technique will be demonstrated on five vowels in terms of visualization and acoustic responses. [This research was supported by NIH R01 CA133015.]

Analysis of high-frequency energy in singing and speech

The human singing and speech spectrum includes energy above 5 kHz, but this portion of the spectrum is typically ignored in speech and voice science. Generally it has been assumed that this high-frequency energy (HFE) contributes to only qualitative percepts of singing and speech, but prior work shows HFE contributes to several non-qualitative percepts, including speech intelligibility. To begin an in-depth exploration of HFE, a database of multi-channel anechoic high-fidelity recordings of singers and talkers was created and analyzed. Third-octave band analysis from the long-term average spectra (LTAS) showed that production level (soft vs. normal vs. loud), production mode (singing vs. speech), and phoneme (for voiceless fricatives) all significantly affected HFE characteristics. Female HFE levels were significantly greater than male levels only above 11 kHz. As expected, HFE was found to be highly directional toward the front of the singer/talker. While this information resulted from a study initially focused on singing voice aesthetic, it is pertinent to various areas of acoustics, including vocal tract modeling, voice synthesis, augmentative hearing technology (hearing aids and cochlear implants), cell phone technology, and training/therapy for singing and speech. [Work supported by NIH-NIDCD.]

Education in acoustics and speech science using vocal-tract models

Several vocal-tract models were reviewed, with special focus given to the sliding vocal-tract model [T. Arai, Acoust. Sci. Technol. 27(6), 384-388 (2006)]. All of the models have been shown to be excellent tools for teaching acoustics and speech science to elementary through university level students. The sliding three-tube model is based on Fant's three-tube model [G. Fant, Acoustic Theory of Speech Production (Mouton, The Hague, The Netherlands, 2006)] and consists of a long tube with a slider simulating tongue constriction. In this article, the design of the sliding vocal-tract model was reviewed. Then a science workshop was discussed where children were asked to make their own sliding vocal-tract models using simple materials. It was also discussed how the sliding vocal-tract model compares to our other vocal-tract models, emphasizing how the model can be used to instruct students at higher levels, such as undergraduate and graduate education in acoustics and speech science. Through this discussion the vocal-tract models were shown to be a powerful tool for education in acoustics and speech science for all ages of students.

A Study on the Principle of Sound Production in Human Whistling using Physical Models of the Human Vocal Tract

The principle of sound production in human whistling is not well-known. The purpose of this letter is to study on the principle of sound production in human whistling using physical models of the human vocal tract. This letter shows that the whistling resonant mode is not also Helmholtz resonance, but also air-column resonance.