Home | About | Journals | Submit | Contact Us | Français |

**|**HHS Author Manuscripts**|**PMC2742369

Formats

Article sections

- Abstract
- I. Introduction
- II. Biomechanical Model of the Vocal Fold Lamina Propria
- III. Results
- IV. Discussion
- V. Conclusion
- References

Authors

Related links

J Voice. Author manuscript; available in PMC 2010 May 1.

Published in final edited form as:

Published online 2008 January 11. doi: 10.1016/j.jvoice.2007.09.010

PMCID: PMC2742369

NIHMSID: NIHMS119722

Corresponding Author: Thomas Siegmund, School of Mechanical Engineering, Purdue University, 585 Purdue Mall, West Lafayette, IN 47907, phone: 765-494-9766, FAX: 765-494-0539, email: ude.eudrup@dnumgeis

This study examines the local and global changes of fundamental frequency (*F*_{0}) during phonation and proposes a biomechanical model of predictions of *F*_{0} contours based on the mechanics of vibration of vocal fold lamina propria. The biomechanical model integrates the constitutive description of the tissue mechanical response with a structural model of beam vibration. The constitutive model accounts for the nonlinear and time dependent response of the vocal fold cover and the vocal ligament. The structural model of the vocal fold lamina propria is based on a composite beam model with axial stress. Results show that local fluctuations such as *F*_{0} overshoots and undershoots can be characterized by the biomechanical model and might be related to the processes of stress relaxation of vocal fold tissues during length changes. The global changes of *F*_{0} declination in declarative sentence production can also be characterized by the model. Such *F*_{0} declination is partially attributed to the peak stress decay associated with stress relaxation of the vocal fold lamina propria and partially to neuromuscular control of the vocal fold length.

The fundamental frequency (*F*_{0}) is a key characteristic of human phonation, which is produced by vibration of the vocal fold when air flows through the glottis. Extensive efforts have been made in signal processing to describe and predict the temporal contour of *F*_{0} changes during speech or singing, e.g., Saitou et al [1] proposed a *F*_{0} control model and showed several types of *F*_{0} fluctuations in singing voice including pronounced over and undershoots at pitch changes; Kutik et al [2] reported a tendency for the fundamental frequency to drop towards the end of a declarative sentence, which is termed *F*_{0} declination; Möbius [3] reported over and undershoots associated with syllables. However, few studies have been conducted to explain these local fluctuations and global changes of *F*_{0} from a predictive point of view. In this study, a biomechanical model of the vocal fold lamina propria is proposed and used to predict the local and global changes of *F*_{0} during speech. This model might provide a better understanding of the mechanisms underlying these phenomena, as opposed to signal processing models which focus on only the description of *F*_{0} changes during phonation.

The biomechanical model of phonation starts with structural models of vocal fold vibration. The ideal string model [4] is the simplest model and is commonly used to predict *F*_{0}. e.g. [5]. However, the string model tends to underpredict the fundamental frequency [6], since the bending stiffness of the vocal fold lamina propria is neglected. Various beam models [6-9] have been proposed to characterize the vibration of the vocal fold. Titze and Hunter [7] developed a beam model, which considered the vocal ligament as the dominant load-carrying component of the lamina propria and idealized the tissue mechanical response as linear elastic. Descout et al [8] developed a multi-layer beam model also with linear elastic properties. Bickley [9] considered the vocal fold as a homogeneous beam without accounting for the effects of vocal fold stretch. A two-layer composite beam model of the vocal fold lamina propria is described in our previous study [6]. This model incorporates both the vocal fold cover and the vocal ligament. It accounts for vibration in the medial-lateral direction (i.e., the predominant direction of vocal fold vibration). Numerical solutions to the beam vibration equations under consideration of axial stress were well approximated by the sum of the closed form solution of the beam model for the unstretched state and the solution of the string model, both dependent on the geometrical properties of the vocal fold lamina propria and also the tissue mechanical response under stretch. To incorporate an accurate description of the vocal fold lamina propria, a constitutive model which can reliably characterize the mechanical response of vocal fold tissue must be utilized. Experimental studies [10, 11] have shown that vocal fold tissues, like other biological soft tissues, demonstrate both elastic and viscous mechanical behavior which is characterized by nonlinearity, hysteresis and time dependence of the stress-stretch response under cyclic deformation.

Extensive studies have been conducted on the mechanical properties of the vocal fold lamina propria. Min et al [12] proposed a nonlinear elastic model which captured the equilibrium stress-stretch response but failed to describe the hysteresis of the tissue. Viscoelastic models were employed by Hunter et al [13] and Chan et al [14, 15] to simulate the response of the vocal fold lamina propria under cyclic tensile or shear deformation. Our previous work [6] documents a two-network constitutive model to describe the stress-stretch response of the vocal fold lamina propria. The two-network model, which consists of a time-independent equilibrium network based on the Ogden's model of hyperelasticity [16] in parallel with a viscoplastic network, proved effective in characterizing the stress-stretch response of the vocal fold lamina propria [11]. The two-network model can describe the hysteretic response in individual load cycles but does not capture long-term viscous behavior. For example, during cyclic stretching and releasing of vocal fold tissues (a condition also termed “preconditioning”), the corresponding level of stress decreases with the number of cycle. Previous studies on brain tissues [17] indicate that such a transient response can be approximated by use of a double-exponential decay function.

The viscous response exhibited by the vocal fold tissues may help explain some phenomena in voice production. It is hypothesized that transient and time-dependent events in the fundamental frequency (over and undershoots as well as declinations) can be partially attributed to the time-dependent response of the vocal fold cover and the vocal ligament represented through the proposed constitutive model of the vocal fold tissue.

In this study, the two-layer composite beam model of vocal fold vibration [6] will be combined with the two-network constitutive model of the vocal fold lamina propria [11], with the constitutive model simulating the mechanical response and the structural model offering the predictions of *F*_{0} under given stretch histories. These stretch histories are representative of transient changes of vocal fold length during phonation [18]. Length changes of the vocal fold are known to control fundamental frequencies and are driven by the activities of intrinsic laryngeal muscles, primarily the cricothyroid (CT) and the thyroarytenoid (TA) muscles [19]. Temporal correlation of muscle activation with syllable formation has also been documented [20]. To exemplify the approach, the constitutive model is applied to describe the mechanical response of cadaveric vocal fold tissues obtained from a 66-year-old male subject. With the tissue constitutive parameters determined, we employ the beam model for predictions of fundamental frequencies for several stretch histories each accounting for the contributions of active (neuromuscular) control.

Vocal fold cover and vocal ligament were dissected from a 66-year-old male larynx excised 15 hours postmortem [10]. This larynx was used for the modeling throughout the present study. The subject was a nonsmoker and Caucasian, and free of head and neck disease and laryngeal pathologies. Experimental data on viscoelastic response of the vocal fold cover and the vocal ligament were obtained by use of a dual-mode servo control lever system (Aurora Scientific Model 300B-LR, Aurora, ON, Canada) with details given in Chan et al. [10].

The network model used to characterize the mechanical response of vocal fold tissue is motivated by the molecular structure of the vocal fold extracellular matrix (ECM). It is assumed that the fibrous proteins (collagen and elastin) provide a structural framework and a large part of the stiffness, whereas the interstitial proteins contribute to regulate the viscosity and elasticity of the ECM. A constitutive model, which consists of a time-independent equilibrium network (A) in parallel with a time-dependent network (B), is thus proposed to reflect the molecular structure of the vocal fold ECM. Following this parallel arrangement of networks, the applied deformation on the whole structure is identical to the deformation in each network, whereas the total stress is the sum of the stresses in these two networks. The time-independent equilibrium network is characterized by Ogden's hyperelastic model which provides a nonlinear correlation between stretch and stress. For uniaxial loading conditions with applied stretch *λ _{u}* the corresponding Cauchy stress

$${\sigma}_{A}=\frac{2{\mu}_{A}}{\alpha}\left({\lambda}_{u}^{\alpha}-{\lambda}_{u}^{-\alpha /2}\right)$$

(1)

where *σ _{A}* is the Cauchy stress and

The second network describes a decomposition of deformation into a hyperelastic component (again described by the Ogden model) in series with a viscoplastic component characterized by a flow rule proposed by Bergström and Boyce [21]. The rate of equivalent plastic stretch is given by:

$${\stackrel{\dot{\xaf}}{\lambda}}_{B}^{\text{pl}}=Z{\left({\overline{\lambda}}_{B}^{\text{pl}}-1+\delta \right)}^{c}{\left({\overline{\sigma}}_{B}\right)}^{m}$$

(2)

in which
${\overline{\lambda}}_{B}^{p}$ is the effective inelastic stretch, *m* is a stress exponent characterizing the dependence of the inelastic deformation on the stress level in network (B), *c* is a stretch exponent characterizing the dependence of the rate of inelastic deformation on the current magnitude of inelastic deformation, *Z* is a viscosity scaling constant defining the absolute magnitude of the inelastic deformation, and *δ* is a small positive number (*δ* < 0.001) introduced to avoid singularities in the inelastic stretch rate when the inelastic stretch is close to unity. The procedure to determine the values of the constitutive parameters from the first cycle of the experimental stress-stretch curves is given in detail in Zhang et al. [11].

The peak stress values recorded during cyclic loading can be approximated by exponential decay functions of the form

$${\sigma}_{\text{p}}\left(t\right)={\sigma}_{\text{0}}\phantom{\rule{0.2em}{0ex}}\left[{a}_{\text{s}}+\sum _{k=1}^{2}{a}_{k}\cdot exp\left(-t/{\tau}_{k}\right)\right]$$

(3)

in which *t* is time, *σ*_{0} is the value of the peak stress in the first cycle, (*σ*_{0}*a*_{s}) is the predicted stabilized peak stress level, and *τ _{k}* (

$${\mu}_{A}\left(t\right)={\mu}_{A,0}\phantom{\rule{0.2em}{0ex}}\left[{a}_{\text{s}}+\sum _{k=1}^{2}{a}_{k}\cdot exp\left(-t/{\tau}_{k}\right)\right]$$

(4)

with *μ _{A}*

The structural model of vocal fold vibration is motivated by the layered structure of the vocal fold lamina propria which is composed of the vocal fold cover (i.e., the epithelium and the superficial layer of the lamina propria) and the vocal ligament (i.e., the intermediate and the deep layer of the lamina propria). These two layers possess different geometrical (shape and cross-sectional area) and mechanical properties (stiffness and degree of nonlinearity) [5, 10]. In the two-layer composite beam model [6], the cross-section of the lamina propria is approximated by two adjacent rectangles that represent the cover and the ligament, respectively. The line connecting the centroid positions of the two rectangles is in the medial-lateral direction. Under the assumption of incompressible material response of vocal fold tissues, changes in cross-sectional areas with tensile deformation are accounted for. When the composite beam is under tension the partial differential equation governing its vibration is

$$\overline{\sigma}\frac{{\partial}^{2}v}{\partial {x}^{2}}-{\overline{E}}_{0}{\kappa}^{2}\frac{{\partial}^{4}v}{\partial {x}^{4}}=\rho \frac{{\partial}^{2}v}{\partial {t}^{2}}$$

(5)

where *v* is the medial-lateral displacement from the equilibrium position of an infinitesimal part of the beam with anterior-posterior distance *x* from one end of the beam, is the homogenized Cauchy stress in the anterior-posterior direction, *ρ* is the tissue density, and *Ē*_{0}*κ*^{2} is a measure of the bending stiffness for the composite beam with *Ē*_{0} being the homogenized anterior-posterior modulus of the lamina propria and κ the radius of gyration of the composite cross-section. With clamped boundary conditions assumed at the vocal process (the posterior endpoint *x* = 0) and at the anterior commissure (the anterior endpoint *x* = *l*), the fundamental frequency of the composite beam can be calculated analytically for the unstretched and the stretched state separately. Furthermore, numerically predicted values of *F*_{0} from Eq. (5) agreed very well with the two closed form analytical solutions in their respective ranges of validity [6]. Over a wide range of vocal fold stretch relevant to phonation, the numerical *F*_{0} solution can be approximated by the sum of *F*_{0} values as predicted from the well known string model and the beam model in the unstretched state

$${F}_{0}\approx \frac{1}{2L{\lambda}_{u}}\sqrt{\frac{\overline{\sigma}}{\rho}}+\frac{3.5607}{{L}^{2}}\sqrt{\frac{{\overline{E}}_{0}{\kappa}^{2}}{\rho}}$$

(6)

where *L* is the *in situ* length (i.e., the stress-free length of the vocal fold), and *λ _{u}* is the longitudinal stretch applied on the vocal fold. Through the ratio between the areas of the cover and the ligament, the homogenized stress in Eq. (6) is related to the stress in the vocal fold cover

The *in situ* length of the vocal fold dissected from the 66-year-old male subject was *L* = 20.425 mm, the initial cross-sectional area of the vocal fold cover was *A*_{0,}* _{c}* = 0.098 cm

The first cycle of the stress-stretch curves was characterized by the two-network constitutive model. The following parameters characterize the constitutive response of the cover specimen, *μ _{A}*

Comparison between experimental data and simulation results of the first cycle of the tensile stress-stretch response at 1 Hz for the vocal fold cover and the vocal ligament of the 66-year-old male subject.

Comparison between experimental data and simulation results by the improved constitutive model of the 15th cycle of the tensile stress-stretch response at 1 Hz for the vocal fold cover and the vocal ligament of the 66-year-old male subject.

Figure 1 compares the experimental stress-stretch curve of the cover and the ligament to the constitutive model characterization. It can be seen from Fig. 1 that the nonlinear and hysteretic stress-stretch curves can be characterized very well by the two-network constitutive model both for the cover and the ligament specimen. Figure 2 exhibits the dependence of the experimentally determined peak values of stress *σ*_{p} for the vocal fold cover and the vocal ligament specimens upon the number of load cycles (or time). Figure 2 also shows the exponential fits obtained through curve-fitting for the cover and the ligament. Normalizing these fits by the values of peak stresses at *t* = 0 (72.3 kPa for the vocal fold cover and 59.6 kPa for the vocal ligament) yielded the long-term decay functions for the cover, *σ*_{p,cover}(*t*)/72.3 = 0.68 + 0.32·exp(−*t*/7.37), and the ligament, *σ*_{p,ligament} (*t*)/59.6 = 0.92 + 0.02·exp(−*t*/1.1) + 0.07·exp(−*t*/8.6). The stiffness of the time-independent equilibrium network evaluated from the first cycle of the stress-stretch curves was *μ _{A}*

Comparison between experimental data and exponential fits of the peak stress decay with time for the vocal fold cover and the vocal ligament of the 66-year-old male subject.

The fundamental frequency response under prescribed transient stretch histories was simulated with the data of the two-network constitutive model. Figure 4(a) depicts a transient stretch history with an initial increase in stretch and a subsequent square wave form under constant mean value of stretch. Figure 4(b) and (c) display the predicted *F*_{0} contours under the conditions of (*i*) *μ _{A}* =

(a) Stretch history under which *F*_{0} contours are predicted; (b) *F*_{0} contour predicted for the 66-year-old male subject with the original constitutive model employed; (c) *F*_{0} contour predicted for the 66-year-old male subject with the improved constitutive **...**

The objective of the present study is to propose a biomechanical model of the vocal fold lamina propria that predicts the local fluctuations and global changes of *F*_{0} during phonation, under given stretch histories. These changes have been documented extensively [1-3] using signal processing methods to describe temporal *F*_{0} contours. This paper aims to describe these changes from a different perspective. The biomechanical model of phonation is based upon a constitutive model for vocal fold tissues [11] and a structural model of vocal fold vibration [6] in our previous studies.

Experimental stress-stretch data were obtained from the vocal fold cover and the vocal ligament [10]. The constitutive model was utilized to characterize the large-strain, hyperelastic, and viscous response of the tissue components under tensile deformation in the anterior-posterior direction. The stiffness of the equilibrium network was prescribed to decrease synchronously with the peak stress to capture the peak stress decay observed in the preconditioning of vocal fold tissues. As can be seen from Figs. 1 and and3,3, the two-network model including a description of the stiffness decay in the equilibrium network can characterize experimental stress-stretch curves very well.

The model was employed to predict *F*_{0} contours under different vocal fold stretch histories. Stretch histories are the active driving parameters in the present study. Real stretch histories in phonation are functions of the antagonistic activities of the cricothyroid (CT) muscle and the thyroarytenoid (TA) muscle. Vocal fold lengthening is primarily attributed to CT activation [5, 20]. It must be noted that the stretch histories implemented were hypothetical but reasonable since they were based upon previous studies [6, 18, 22-25] on the length changes of vocal fold during phonation. It can be seen from Figs. 4 and and55 that overshoots and undershoots of *F*_{0} [1] (i.e., the local fluctuations *F*_{0}) can be characterized by our model corresponding to the step activation of stretch followed by hold periods in the stretch histories. The present model predicts that the overshoot and undershoot response is caused by stress relaxation in the vocal fold lamina propria. When the singer (or speaker) lengthens or shortens the vocal fold, and then keeps it at a constant length for a period of time, the stress in the vocal fold relaxes and causes the *F*_{0} to drop or rise depending upon whether the vocal fold is lengthened or shortened. The global changes of *F*_{0} are also characterized with the present biomechanical model of phonation. Under constant mean value of stretch the decay percentage of peak *F*_{0} predicted in Fig. 4(c) is not enough compared to the actual decay percentage characteristic of declination in declarative utterances. This appears to indicate that the viscous response of the vocal fold lamina propria can only contribute partially to the fundamental frequency decay phenomenon in declarative utterances and that active linguistic and neuromuscular control of vocal fold stretch also plays a significant role in the formation of a declination.

The present constitutive model is not capable of characterizing some of the other phenomena associated with cyclic loading of vocal fold tissue (e.g., the gradual increase of the *in situ* (stress-free) length). A three-network constitutive model with two characteristic time constants has to be formulated to comprehensively characterize not only the short-term but also the long-term mechanical behaviors of the vocal fold lamina propria. Such a model is currently under investigation. In this sense, the current two-network model with an imposed exponential decay of the stiffness of the hyperelastic network is an intermediate model between the two-network model and the comprehensive three-network model. Limitations of the biomechanical model also originate from the structural model of vocal fold vibration, which only accounts for vibration in the medial-lateral direction. Physiologically, vocal fold vibration also involves motion in the inferior-superior direction. In this direction the vocal fold as a structure is more resistive than in the medial-lateral direction. Hence, the real *F*_{0} of vocal fold vibration is probably higher than those predicted from the medial-lateral direction and thus much more complex to analyze.

The present study proposes a biomechanical model of phonation whose driving parameter is the stretch history as related to muscle activity. Since the model is based on parameters that describe actual tissue specimens it is in principle capable of predicting phonation differences between individual speakers. The contractions of the CT muscle and the TA muscle play antagonistic roles in regulating the vocal fold length changes. The vocal fold stretch histories employed in this study shift among different levels, which correspond to different syllables. The model demonstrates that the global (declination) and local changes (overshoot and undershoot) of fundamental frequency during phonation can in part be predicted by this biomechanical model as an outcome of the viscoplastic response of the tissue. Fundamental frequency changes over the time span of an utterance, however, appear to require additional active neuromuscular control of the vocal fold length. Such a model could be built by combining the present results with models involving active control, such as a body-cover model [5, 22].

This work was supported by the National Institute on Deafness and Other Communication Disorders, NIH grant no. R01 DC006101.

**Publisher's Disclaimer: **This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1. Saitou T, Unoki M, Akagi M. Development of an *F*_{0} control model based on *F*_{0} dynamic characteristics for singing-voice synthesis. Speech Commun. 2005;46:405–417.

2. Kutik EJ, Cooper WE. Declination of fundamental frequency in speakers' production of parenthetical and main clauses. J Acoust Soc Am. 1983;73:1731–1738. [PubMed]

3. Möbius B. Components of a quantitative model of German intonation. Proceedings of the 13th International Congress of Phonetic Sciences; Stockholm. 1995. pp. 108–115.

4. Rayleigh . The Theory of Sound. Vol. 1. The Mcmillan Company; reprinted 1945 by Dover Publications, New York.

5. Titze IR. Principles of Voice Production. Englewood Cliffs, NJ: Prentice-Hall; 1994.

6. Zhang K, Siegmund T, Chan RW. A two-layer composite model of the vocal fold lamina propria for fundamental frequency regulation. J Acoust Soc Am. 2007;122:1090–1101. [PubMed]

7. Titze IR, Hunter EJ. Normal vibration frequencies of the vocal ligament. J Acoust Soc Am. 2004;115:2264–2269. [PMC free article] [PubMed]

8. Descout R, Auloge JY, Guerin B. Continuous model of the vocal source. International Conference on Acoustics, Speech and Signal Processing; Denver, CO. 1980. pp. 61–64.

9. Bickley C. PhD dissertation. Massachusetts Institute of Technology; Cambridge, MA: 1989. Acoustic evidence for the development of speech.

10. Chan RW, Fu M, Young L, Tirunagari N. Relative contributions of collagen and elastin to elasticity of the vocal fold under tension. Ann Biomed Eng. 2007;35:1471–1483. [PubMed]

11. Zhang K, Siegmund TH, Chan RW. A constitutive model of the human vocal fold cover for fundamental frequency regulation. J Acoust Soc Am. 2006;119:1050–1062. [PubMed]

12. Min YB, Titze IR, Alipour-Haghighi F. Stress-strain response of the human vocal ligament. Ann Otol Rhinol Laryngol. 1995;104:563–569. [PubMed]

13. Hunter EJ, Titze IR, Alipour F. A three-dimensional model of vocal fold abduction/adduction. J Acoust Soc Am. 2004;115:1747–1759. [PMC free article] [PubMed]

14. Chan RW, Siegmund T. Vocal fold tissue failure: preliminary data and constitutive modeling. J Biomech Eng. 2004;126:466–474. [PubMed]

15. Chan RW, Titze IR. Viscoelastic shear properties of human vocal fold mucosa: Theoretical characterization based on constitutive modeling. J Acoust Soc Am. 2000;107:565–580. [PubMed]

16. Ogden RW. Larger deformation isotropic elasticity – on the correlation of theory and experiment for incompressible rubberlike solids. Proc Royal Soc London, Series A. 1972;326:565–584.

17. Gefen A, Margulies SS. Are in vivo and in situ brain tissues mechanically similar? J Biomech. 2004;37:1339–1352. [PubMed]

18. Hollien H. Vocal pitch variation related to changes in vocal fold length. J Speech Hear Res. 1960;3:150–156.

19. Hirano M, Ohala J, Vennard W. The function of laryngeal muscles in regulating fundamental frequency and intensity of phonation. J Speech Hear Res. 1969;12:616–628. [PubMed]

20. Poletto CJ, Verdun LP, Strominger R, Ludlow CL. Correspondence between laryngeal vocal fold movement and muscle activity during speech and nonspeech gestures. J Appl Physiol. 2004;97:858–866. [PMC free article] [PubMed]

21. Bergström JS, Boyce MC. Constitutive modeling of the time-dependent and cyclic loading of elastomers and application to soft biological tissues. Mech Mater. 2001;33:523–530.

22. Titze IR. The Myoelastic Aerodynamic Theory of Phonation. Iowa City, IA: National Center for Voice and Speech; 2006.

23. Ohala J, Ewan W. Speed of pitch change. J Acoust Soc Am. 1973;53:345A.

24. Sundberg J. Maximum speed of pitch changes in singers and untrained subjects. J Phonetics. 1979;7:71–79.

25. Titze IR, Jiang JJ, Lin E. The dynamics of length change in canine vocal folds. J Voice. 1997;11:267–276. [PubMed]

PubMed Central Canada is a service of the Canadian Institutes of Health Research (CIHR) working in partnership with the National Research Council's national science library in cooperation with the National Center for Biotechnology Information at the U.S. National Library of Medicine(NCBI/NLM). It includes content provided to the PubMed Central International archive by participating publishers. |