Search

A modal analysis is made of a biomechanical simulation of the vocal folds. The dominant modes of the finite element simulation are computed by performing a principal components analysis on the time‐varying nodal coordinates of the model. For simulations of normal phonation, the model−which contains many degrees of freedom−is reduced to essentially two dominant modes corresponding to those predicted by Titze and Strong [‘‘Normal modes in vocal cord tissues,’’ J. Acoust. Soc. Am. 57, 736–744 (1975)]. Additional modes are needed to explain simulations of pathological voice. The dominant modes from the various simulations will be displayed and discussed. The ability of traditional vocal fold models to capture these modes will also be considered. [This research was supported by Grant No. P60 DC00976‐0X from the National Institutes on Deafness and Other Communication Disorders.]

Examination of surface wave properties in the vocal fold mucosa is becoming an important part of assessment of vocal function. A key wave property is propagation velocity, which determines the phase delay between upper and lower margins of the vocal folds. Excised canine larynges were used in this study to measure this phase delay, and therewith propagation velocity. The motion of two fleshpoints was tracked stroboscopically. Differential displacements between the fleshpoints were matched to displacements of a model. A least‐squared fit of the data to the model provided the numerical values of propagation velocity, which varied from 0.5 m/s to about 2.0 m/s, depending on fundamental frequency. The corresponding phase delay along the medial surface of the vocal folds varied from about 60°/mm to 30°/mm. [Work supported by NIDCD, Grant No. P60‐DC0976.]

Vocal intensity is studied as a function of fundamental frequency and lung pressure. A combination of analytical and empirical models is used to predict sound pressure levels from glottal waveforms of five professional tenors and twenty five normal control subjects. The glottal waveforms were obtained by inverse filtering the mouth flow. Empirical models describe features of the glottal flow waveform (peak flow, peak flow derivative, open quotient, and speed quotient) in terms of lung pressure and phonation threshold pressure, a key variable that incorporates the Fo dependence of many of the features of the glottal flow. The analytical model describes the contributions to sound pressure levels SPL by the vocal tract. Results show that SPL increases with Fo at a rate of 8-9 dB/octave provided that lung pressure is raised proportional to phonation threshold pressure. The SPL also increases at a rate of 8-9 dB per doubling of excess pressure over threshold, a new quantity that assumes considerable importance in vocal intensity calculations. For the same excess pressure over threshold, the professional tenors produced 10-12 dB greater intensity than the male nonsingers, primarily because their peak airflow was much higher for the same pressure. A simple set of rules is devised for predicting SPL from source waveforms.

The accuracies of four pitch extraction methods are introduced and compared for high‐precision perturbation measurement. They include combinations of peak‐picking, zero‐crossing, and waveform‐matching strategies. The effects of sampling frequency, signal‐to‐noise ratio, pitch frequency, and signal modulation on the accuracies are explored. It is shown that methods with the waveform‐matching technique give the best results for most cases. For near‐sinusoidal stimuli (e.g., EGG signals), the zero‐crossing technique gives higher accuracy than the peak‐picking technique. At a low signal‐to‐noise ratio, a higher sampling frequency does give better results for the waveform‐matching technique, but the accuracy of the peak‐picking method may reduce as sampling frequency increases. As the signal‐to‐noise ratio increases, the effect of sampling frequency on accuracy diminishes when interpolation is used. A variety of waveforms are currently being tested to check the suitability of these methods for different applications. [Work supported by NIH.]

Electroglottography (EGG) signals arc routinely used for observation of vocal fold adduction and regularity in vocal fold vibration. Analysis over many cycles can be very tedious if waveform parameters arc extracted manually or semiautomatically. A parameter optimization strategy is used here to provide a list fit to a model, thereby allowing larger volumes of data to be processed. Processing speed and reliability of extraction depend on the type of model being used and the cleverness in choosing the appropriate seed values. [Work supported by NIH.]

Summary The abduction quotient, a measure of effective glottal width, was obtained for electroglottographic recordings from a professional operatic baritone singer. The subject produced repeated tokens of the voice qualities breathy, normal, and pressed (or constricted) in both a speech and a singing manner. In the singing manner, the subject produced the three vocal qualities at three pitch levels and three loudness levels. The abduction quotient decreased from breathy to pressed voice, suggesting that the measure corresponds to effective glottal width. The measure was found to be consistently low during all conditions of singing, suggesting that the subject produced all singing tokens with relatively strong laryngeal adduction at the vocal process level. Although the results of this study support the validity and usefulness of the abduction quotient, further verification is needed.

Damage to the vocal cords can result in scarring and impaired vibration and can manifest clinically as hoarseness and loss of vocal power. If the vibratory characteristics could be restored in these scarred vocal cords, the vocal intensity and efficiency of phonation also should improve. In an effort to enhance the vibration of damaged vocal cords, we implanted a submucosal fat autograft within the injured vocal cord cover layer of dogs 6 weeks after unilateral mucosal excision had been performed. Three months postoperatively these animals were compared to normal dogs and those with mucosal excision but no fat-grafting. Acoustic and biomechanical measures of phonation were collected from an excised larynx preparation. We found that the fat-augmented vocal cords had lower threshold pressures for phonation, greater vocal intensity, and more efficient acoustic output than injured vocal cords without the fat-grafting. These results provide a foundation for further research on reconstructive surgery of damaged vocal cords.

Analysis of vocal fold vibration requires information on the viscoelastic properties of the vocalis muscle. The force response of two canine vocalis muscles was measured in one-dimensional, stepwise elongation of the tissue as a function of time with a computer-controlled ergometer. The viscoelastic behavior of the muscle in its passive state was demonstrated through the relaxation test. A quasilinear viscoelastic model was used to parametrize the relaxation function, and results are reported for various levels of strain. Furthermore, a model was used to obtain theoretical time-dependent stress-strain curves to compare with experimental data.

This paper presents a rationale for acoustic analysis of voices of neurologically diseased patients, and reports preliminary data from patients with myotonic dystrophy, Huntington's disease, Parkinson's disease, and amyotrophic lateral sclerosis, as well as from individuals at risk for Huntington's disease. Noninvasive acoustic analysis may be of clinical value to the otolaryngologist, neurologist, and speech pathologist for early and differential diagnosis and for documenting disease progression in these various neurologic disorders.

Summary Archives of voice recordings could provide a source of information for voice scientists and clinicians interested in voice analysis and in cultural and historical changes of vocal styles and qualities. Large collections of sound recordings exist that could be used as databases for such inquiries, although no comprehensive list of such collections currently exists. A preliminary survey of major collections is presented. Voice scientists and voice archivists have many common areas of interest in the recording, reproduction, copying, and analysis of sound. The Recording and Research Center, as part of its ongoing research in establishment of standards for voice analysis, has begun an archive of laboratory recordings. A protocol for recording trained, normal, abnormal, and neurologically impaired voices is in development. This survey of sound archives will continue as well.

The human vocal cord is treated as an elastic material capable of propagating compressional, shear, and surface waves. A mathematical formulation for commonly observed vibrational modes is developed. This includes horizontal as well as vertical tissue displacements. Current vocal cord models are discussed and evaluated in terms of their ability to describe adequately the vibrational degrees of freedom of the cords.Subject Classification: 70.20.

Summary Acoustic and glottographic measures may provide important information that could enhance clinical management and documentation of vocal dysfunction. Acoustic measures such as jitter and shimmer reflect “short-term” perturbations, or instabilities of the voice, and the coefficients of variation for frequency and for amplitude reflect “long-term” perturbations. Interpretations of these acoustic measures are based on the assumption that vocal perturbations may be related to laryngeal tissue abnormalities, asymmetries in vocal fold movement, or neuromuscular fluctuations in the respiratory, laryngeal, or vocal tract systems. The abduction quotient is a glottographic measure related to laryngeal adduction and is obtained from an analysis of the electroglottograph signal. The adduction measure appears to be independent of the acoustic perturbation measures. Interpretations of the acoustic and adductory measures may, therefore, complement each other for greater understanding of a patient's laryngeal behavior. Visual displays of the acoustic and glottographic signals also are discussed to demonstrate their value in voice signal interpretations. Case studies illustrate potential interpretations of the acoustic perturbation and abduction quotient measures.

Summary Register transitions are divided into two classes, periodicity transitions and timbre transitions. Periodicity transitions refer to changes in vocal quality that occur whenever glottal pulses are perceived as individual events rather than as a continuous auditory stimulus. Timbre transitions refer to changes in vocal quality associated with changes in spectral balance. Physiologically, these can be quantified with an abduction quotient. The singing registers appear to be based on timbre transitions resulting from subglottal resonances that interfere with the vocal fold driving pressure. Four of the major singing register shifts are predicted (in frequency and relative importance) on the basis of the first subglottal formant. Strategies for register equalization are proposed on the basis of supraglottal formant tuning (vowel modification) and adjustments in glottal adduction.

The myoelastic-aerodynamic theory of phonation has been quantified and tested with mathematical models. The models suggest that vocal fold oscillation is produced as a result of asymmetric forcing functions over closing and opening portions of the glottal cycle. For nearly uniform tissue displacements, as in falsetto voice, the asymmetry in the driving forces can result from the inertia of the air moving through the glottis. This inertia can in turn be enhanced or suppressed by supraglottal or subglottal vocal tract coupling. More obvious and pronounced asymmetries in the driving forces are associated with non-uniform vocal fold tissue displacements. These are combinations of normal tissue modes, and can result in vertical and horizontal phase differences along the surfaces, as observed in chest voice. The ranges of oscillation increase among various models as more freedom in the simulated tissue movement is incorporated. Of particular significance in initiating and maintaining oscillation are the vertical motions that facilitate coupling of aerodynamic energy into the tissues and allow tissue deformations under conditions of incompressibility. Vertical displacements also can have a significant effect on vocal tract excitation. Control of fundamental frequency of oscillation (FO) is basically myoelastic, partially as a result of deliberate or reflex adjustments of laryngeal muscles, and partially as a result of nonlinear tissue strain over the vibrational cycle. This places limits on the control of FO by subglottal pressure, and forces such control to be inseparably connected with vibrational amplitude, or less directly, with vocal intensity.

The purpose of this investigation was to develop an in vitro technique for measurement of elastic properties of isolated vocal fold tissue. Larynges were excised from anesthetized dogs, immediately submerged in a Krebs-Ringer solution, and aerated with 95% oxygen and 5% carbon dioxide. The tissue was maintained in the aerated electrolyte solution throughout the experiment. The vocalis muscle was carefully dissected, with attachments to the arytenoid and thyroid cartilages maintained. The preparation was then subjected to isometric (constant length) force-elongation measures, which were converted to stress-strain values. Stage I of the investigation identified the dog model with the least intrastudy variability relative to the age, sex, and breed of the research animal, as well as to the effects of curare. Stage II investigated the effects of different instrumentation, definitions of reference length, methods of elongation, and effects of electrical stimulation. Once the procedure had been refined, the effects of age and sex were retested. There was a significant interaction between sex and strain and between age and strain. The least variability was obtained with curarized tissue from one sex of young, mixed-breed dogs, where an arbitrary 1 gram of initial force was the criterion for establishing "zero" strain. Problems associated with determination of reference length and various approaches to this problem are discussed.

Summary This study focuses on the perceptual nature of chest and falsetto registers as a function of various production tokens and methods of perceptual evaluation. Fifteen target tones, ranging from G#3 to A#4, were sung by a male and a female subject in the context of ascending and descending sequences on the vowels /a/ and /i/. Register transitions were elicited by setting strict constraints on production. Segments of 1-s duration were extracted from the target notes, digitized, and acoustically analyzed. These excerpts were presented to ten trained listeners in four different perceptual tasks. Identification and discrimination tasks yielded nearly identical results, suggesting that the primary registers are perceived as distinct entities. The marked change from chest to falsetto as well as the locus of the transition between these registers did not vary systematically as a function of production token or perceptual task. Mean register shift-point frequencies of the male and female subjects were perceived at 320 Hz and 353 Hz, respectively. Multidimensional scaling and hierarchical clustering analyses were utilized to capture the dimensionality and the internal structure of perceptual data sets derived from the pair-wise similarity ratings. Optimal spatial representation of these data required no more than two orthogonal dimensions, with the quality attribute represented by the dominant dimension. The representation of pitch differences was reflected only in the internal ordering of the stimuli within registers, but did not affect the perceptual discontinuity between registers.

The power dissipated in the vocal folds during phonation results from viscous loss in the fold tissues and can be estimated on the basis of a simplified representation of the geometry and movement of the folds. Assuming this power is converted into heat, it combines with the effects of muscle contraction, blood perfusion, and thermal conduction to the surface of the folds to determine fold temperature. To isolate the thermomechanical source of heat, the vocal folds of excised larynges were vibrated vertically over a range of frequencies and amplitudes, while the temperature rise in them was measured by inserted fine-wire thermocouples. The temperature rise observed was somewhat smaller than that predicted. This may be partly accounted for by convective and evaporative cooling of the vocal fold as it moved relative to the ambient air. Future studies should consider the time course of heating and vocal fold geometry in more detail.

Summary Changes in vocal fold oscillation threshold pressure were induced in excised canine larynges by experimentally causing fluid movement into and out of the vocal folds. The transport was facilitated by exposing the vocal folds to various osmotic solutions, and it was assumed that changes in hydration caused changes in the internal tissue viscosity. A range of oscillation threshold pressures was measured for each condition of hydration by varying length and glottal width. The oscillation threshold pressure shifted as predicted. Decreased hydration (increased viscosity) raised the threshold of oscillation, and increased hydration (decreased viscosity) lowered the threshold of oscillation. This apparently represents the first in vitro model for the study of the effect of viscosity changes of the internal environment of the vocal folds on phonation.

The human vocal fold is treated as a continuous, transversally isotropic, bounded medium. The theory of viscoelasticity is applied to formulate the stresses, strains, glottal tissue impedance, and propagation characteristics for sinusoidal oscillation within the vocal‐fold tissues. Incompressibility of the tissues leads to a biharmonic differential equation, the solution of which represents commonly observed modes of vibration. The energy dissipated in the tissues and the associated damping factors are calculated. Mode impedances of the tissues are calculated and lumped‐model approximations of the system evaluated. The usual mass‐spring models are shown to be inadequate in certain important cases, especially during vocal‐fold closure and for modes in which vertical phase differences occur.Subject Classification: [43]70.20.

Summary Thyroarytenoid muscle activity is shown to combine with cricothyroid muscle activity to regulate fundamental frequency of phonation. The relative amount of activity in these muscles, as measured electromyographically, is illustrated on a muscle activation plot (MAP) for four subjects vocalizing at different pitches and loudnesses. Electrical stimulation of the thyroarytenoid muscle in various regions of the MAP suggests that both positive and negative changes in fundamental frequency ( F 0 ) can occur with increased thyroarytenoid activity. At lower fundamental frequencies and lower vocal intensities, F 0 correlates positively with thyroarytenoid activity, but at higher fundamental frequencies and low intensity (especially in falsetto voice) an increase in thyroarytenoid activity tends to lower F 0 . A biomechanical body-cover model of fundamental frequency control is used to explain this phenomenon.

Articulatory models of speech production are generally controlled by kinematic parameters that govern the position and movement of tissue and air. A four parameter model of the glottis is described with similar kinematic parameters to complement this approach. Parameters are nondimensional zed quotients that control the static and dynamic shape of the glottis. The model can stimulate glottal flow, glottal area, and vocal contact area simultaneously which identical parameters. Specific features of these waveforms are discussed in terms of the glottal configuration that was used to produce them. In particular, a detailed description of the contact area waveshape is included. The model is expected to be applied to vocal fold imaging techniques that rely on signals gathered noninvasively on the surface of the body. For synthesis, the model provides an alternative to flow pulse modeling because it can include some source-system interactions with relatively little computational overhead.

Summary Preliminary theoretical and empirical work suggest that increased thyroarytenoid muscle activity may either increase or decrease fundamental frequency, depending on cricothyroid muscle activity and a new cross-sectional area parameter. This parameter is defined as the ratio of muscular tissue in vibration to total tissue in vibration. Canine laryngeal nerves were stimulated to measure vocal-fold length changes. These data, combined with previously reported tissue density, passive stress, and passive frequency data, were used to construct a set of curves predicting canine fundamental frequency from thyroarytenoid and cricothyroid muscle activity and the area ratio. The results suggest that high cricothyroid muscle activity and small area ratios tend to cause fundamental frequency lowering with increased thyroarytenoid muscle activity.

Active properties of canine vocalis muscle tissue were investigated through a series of experiments conducted in vitro. Samples of the vocalis muscle were dissected from dog larynges excised a few minutes before death and kept in Krebs solution at a temperature of 37 ± 1°C and a pH of 7.4 ± 0.05. Isometric and isotonic tetanic responses of the vocalis muscle were obtained electronically with a Dual Servo System (ergometer). Isometric tension was recorded at various levels of elongation and stimulation rate. Isotonic shortening was recorded at various levels of force, and shortening velocity was obtained by numerical analysis of recorded data. It was found that fused tetanus occurred at stimulation rates of about 90 Hz, where the isometric titanic force saturates. Repeated stimulation of the muscle in vitro not only caused nonrecoverable fatigue in the tissue, but also decreased its passive tension. The combined active and passive isometric tension increased with elongation of the muscle. Results of isometric active responses were normalized with respect to average passive response. This normalization allowed for better comparison between tetanic contraction and twitch contraction. It was found that maximum tetanic contraction was 6.4 times greater than maximum twitch contraction obtained in a previous study. A tetanic contraction period was defined and investigated for eight samples of vocalis muscle tissue from different dogs. The tetanic contraction period showed a linear increasing trend with strain.

A mathematical model for digital computer simulation of human-like utterances is developed. The overall system consists of a period structure of 16 coupled masses for each of the vocal cords, an 18-section cylindrical tube approximation of the pharynx and mouth, and a similar 12-section nasal tract. Special care has been taken to model separately the functions of the vocal ligament, the vocalic muscle, and the mucosa. Simulated phonation in modal, mixed and falsetto registers is possible. The parameters which control the nature of the phonation are lung pressure, external tension applied to ligament, vocalis, and mucosa, and the internal muscular action of the vocalis. Applications are cited in areas which include vocal physiology, pathology, and pedagogy.

Thomas Shipp: Our panelists will make a few statements, exchange some information and then we will open the discussion to the audience. Professor Sundberg. Johan Sundberg: Larynx height is a fascinating aspect of singing, in a physiological sense, and I would like to add a few comments. One is that I think that most speakers have the habit of raising their larynx with pitch, so that is the starting point for any voice training as far as I know. Larynx height has an acoustical effect on the voice organ, in that it shortens the resonance cavity in the pharynx, and this affects all format frequencies. If you change the format frequency, which you do by raising the larynx, then you change the voice timbre. I have a concern in regards to your findings that glissando was affiliated with larynx rises in professional singers. That would mean that when they sang glissando they would accept a change in voice timbre with pitch, and I don't think that is necessarily esthetically a great thing to do. I think that timbre should change, not for pitch reasons, but for expressive reasons. I would like to raise the question if you think that the effect of the glissando and larynx height was due to the fact that the glissando is a rather lovely practiced thing in singing music. My next point is that the larynx height has an acoust ical effect not only on the resonance chambers of the outer voice organ but also on the source itself. I have always thought that as larynx lowering is associated with inhaling, and inhaling is associated with glottal widening, then there is probably an abductory component in the larynx low

Simulation of glottal volume flow and vocal fold tissue movement was accomplished by numerical solution of a time-dependent boundary value problem, in which nonuniform, orthotropic, linear, incompressible vocal fold tissue media were surrounded by irregularly shaped boundaries, which were either fixed or subject to aerodynamic stresses. Spatial nonuniformity of the tissues was of the layered type, including a mucosal layer, a ligamental layer, and muscular layers. Orthotropy was required to stabilized the vocal folds longitudinally and to accomodate large variations in muscular stress. Incompressibility and vertical motions at the golttis played an important role in producing and sustaining phonation. A nominal configuration for male fundamental speaking pitches was selected, and the regulation of fundamental frequency, intensity, average volume flow, and vocal efficiency was investigated in terms of variations around this nominal configuration. Parameters which were varied consisted of geometrical factors such as length, thickness, and depth, factors for shaping the glottis, as well as tissue elasticities, tissue viscosities, and subglottal pressure. Since nonlinear stress-strain properties were not included, subglottal pressure did not produce a pronounced effect upon fundamental frequency under these somewhat edealized conditions F0 rasing correlated strongly with increased tension in the ligament, and somewhat with increasing tension in the vocalis. F0 lowering correlated with increase in vocal fold length when the tensions were held constant, but not with increase in vocal fold thickness. Vocal intensity and efficiency are shown to have local maxima as the configurational parameters are varied one at a time. It appears that oral acoustic power output and vocal efficiency can be maximized by proper adjustments of longitudinal tension of nonmuscular (mucosal and ligamental) tissue layers in relation to muscular layers. Quantitative verification of the "body-cover" theory is therefore suggested, and several further implications with regard to control of the human larynx are considered.

Comparison is drawn between male and female larynges on the basis of overall size, vocal fold membranous length, elastic properties of tissue, and prephonatory glottal shape. Two scale factors are proposed that are useful for explaining differences in fundamental frequency, sound power, mean airflow, and glottal efficiency. Fundamental frequency is scaled primarily according to the membranous length of the vocal folds (scale factor of 1.6), whereas mean airflow, sound power, glottal efficiency, and amplitude of vibration include another scale factor (1.2) that relates to overall larynx size. Some explanations are given for observed sex differences in glottographic waveforms. In particular, the simulated (computer-modeled) vocal fold contact area is used to infer male-female differences in the shape of the glottis. The female glottis appears to converge more linearly (from bottom to top) than the male glottis, primarily because of medial surface bulging of the male vocal folds.

The maximum theoretical quantization noise (shimmer and jitter) for digital recording systems is estimated and compared with normal vocal shimmer and jitter. Nine bits of resolution and 500 samples per cycle are needed to minimize the contaminating sampling noise without interpolation between samples. With interpolation, however, fewer than 100 samples per cycle can resolve jitter down to 0.1%. Low-pass filtering is not harmful unless peak-picking strategies are used, and the peaks are severely broadened by the low-pass filter. A window of at least 20 cycles is suggested, and multiple tokens of an utterance are necessary to obtain a stable mean for perturbation measures.

The pressure-flow equations used in computer simulation studies of phonation lack experimental validation. Two polyester resin models of the laryngeal airway with rectangular glottal ducts were constructed in order to obtain the relationships between translaryngeal pressure drop and volume flow through the airway. The results are in disagreement with the early estimates of Wegel [Bell Syst. Tech. J. 9, 207-227 (1930)], but match the predictions given by Ishizaka and Matsudaira [SCRL Monograph No. 8 (1972)] to within approximately +/- 10% for typical translaryngeal pressures for speech, with larger discrepancies being found for the model with the larger glottal diameter. The equation given by van den Berg et al. [J. Acoust. Soc. Am. 29, 626-631 (1957)] may not be properly compared because their supraglottal pressure hole location may have been different from that used in the present study. The data from the two models also are compared to recent empirical studies using an enlarged model of the larynx [J. Gauffin et al., Conference on Vocal Fold Physiology, Madison (1981)].

Vocal cord injury and its effect on the larynx are topics of considerable importance, yet a clear understanding of acute vocal cord injury from excessive phonation remains elusive. The inability to develop an animal model for vocal abuse has hampered research in this area. This project describes the development of a canine model for acute vocal cord injury from excessive phonation. The method of phonation consisted of a humidified, temperature-controlled, constant flow of air delivered subglottally by an external compressor. A small group of dogs underwent phonation for 2 and 4 hours. Vocal cord injury is described using histologic techniques. Damage to the vocal cords, other than hemorrhage (rare in dogs) and edema, proved to be too elusive for routine microscopy. Scanning and transmission electron microscopy showed extensive surface damage with destruction and loss of the surface microridges and premature desquamation of the vocal cord squamous epithelium. Injury extended to the lamina propria, where the collagen fibers were detached from the basement membrane.

A new set of parameters is described for analysis and synthesis of glottal area, vocal fold contact area, and glottal volume flow. Parameters are all nondimensionalized and consist of an abduction quotient, a shape quotient, a phase quotient, and a load quotient in addition to fundamental frequency and vibrational amplitude. The parameters show promise in interpretation of electroglottographic, photoglottographic, and inverse filtered volume velocity waveforms in terms of the glottal configuration. Some comparisons between modeled and measured glottographic waveforms are made.

A theory of vocal fold oscillation is developed on the basis of the body-cover hypothesis. The cover is represented by a distributed surface layer that can propagate a mucosal surface wave. Linearization of the surface-wave displacement and velocity, and further small-amplitude approximations, yields closed-form expressions for conditions of oscillation. The theory predicts that the lung pressure required to sustain oscillation, i.e., the oscillation threshold pressure, is reduced by reducing the mucosal wave velocity, by bringing the vocal folds closer together and by reducing the convergence angle in the glottis. The effect of vocal tract acoustic loading is included. It is shown that vocal tract inertance reduces the oscillation threshold pressure, whereas vocal tract resistance increases it. The treatment, which is applicable to falsetto and breathy voice, as well as onset or release of phonation in the absence of vocal fold collision, is harmonized with former treatments based on two-mass models and collapsible tubes.

The twitch response of the canine vocalis muscle was investigated through a series of experiments conducted in vitro. Samples of vocalis muscle were dissected and prepared from canine larynges a few minutes before death and kept in Krebs solution at a temperature of 37 ± I°C and pH of 7.4 ± 0.05. Field stimulation with parallel-plate silver electrodes was applied to study the twitch response of muscle samples. The peak tension and time course of isometric contraction of isolated muscle samples were measured electronically with a Cambridge Technology Dual Servo System (ergometer). Contraction time and 50% relaxation time of this muscle were measured for seven samples at various levels of strain. It was found that contraction time ranged between 22 and 32 ms and 50% relaxation time ranged between 17 and 37 ms. Results indicate that the vocalis muscle is a fast muscle capable of performing rapid maneuvers in support of changes in fundamental frequency.

A small‐oscillation theory is developed for determining threshold conditions for onset and release of vocal fold vibration as a function of lung pressure, tissue viscosity, glottal aperture, and surface wave velocity in the mucosa (vocal fold cover). It is found that oscillation threshold lung pressure increases in proportion to mucosal wave velocity. In other words, phonation initiates at lower pressures (less effort) if the vocal fold cover is lax and propagates slow surface waves. Experimental evidence on excised dog larynges confirms that a stiffer cover (at greater vocal fold lengths) requires greater lung pressure. The theoretical prediction that increased glottal aperture (spreading of the vocal folds, as in breathy voice) also increases the oscillation threshold lung pressure has been more difficult to verify experimentally, but increased tissue viscosity produces an observable increase. Applications to phonetics and voice care are cited. [Work supported by NINCDS.]

Three models of phonation are described, one at the acoustic level, one at the kinematic level, and one at the biomechanic level. The models all feed the same vocal tract, which is the partial wave (Kelly‐Lochbaum) transmission‐line analog. The first source model is noninteractive with the vocal tract, the glottal pulse being computed by formula [Titze, J. Acoust. Soc. Am. Suppl. 1 68, S71 (1980)]. The second model is interactive in terms of the flow, but specifies vocal fold movement using a formula [Titze, J. Acoust. Soc. Am. 75, 570 (1984)]. The third is interactive in terms of flow and tissue movement, i.e., it self‐oscillates. Comparisons are drawn between the three models in terms of quality of sound produced, input parameters specified, and nonauditory outputs derived. Vocal tract area input to the models has recently been modified to be compatible with the Haskins articulatory synthesizer (companion paper in this session), so that sentence‐length utterances can be produced.

This study of vocal fold vibration examines the physics of vocal fold vibration and some possible causes of perturbed vibration. A multiple mass‐spring computer model is used to simulate the vibratory motion of the normal vocal fold and a fold with asymmetric parameters. Speech, displacement, acceleration plots, and a mathematical discussion reducing the model to a function of lateral tissue motion are presented. Acceleration‐displacement (energy exchange) diagrams from the simulation model are compared to the behavior of the mathematical equations, showing that the system driving forces can be associated with physical phenomena such as vocal tract loading, vertical phasing, Bernoulli forces, and negative damping. Simulations have produced perturbed vibrations, suggesting that 1/2 and 1/3 subharmonics are present. The 1/2 subharmonic results from vocal tract resistive and inertive loading, while the 1/3 subharmonic arises due to inertive loading and nonlinear tissue stiffness. It is necessary to derive a ...

Acoustic analyses of the human voice have been developed to assist in clinical screening of laryngeal pathology. It is hypothesized that perturbation functions derived from parameters measured from the acoustic signal might reflect phonational irregularities due to randomness in the action potentials governing muscle tension, asymmetry in vocal fold geometry, randomness in the air stream from the glottis, interactions between source and tract, and other, as yet unknown, factors. A set of measures computed from these perturbation functions could provide a quantitative description of vocal fold pathology. Unfortunately, many perturbation measures in use today were defined in an ad hoc fashion, with little attention paid to comparatively assessing proposed measures against existing measures. This paper represents an attempt to put the study of voice perturbations on a sounder theoretical footing. Utilizing theories of finite differences and subsampling, algorithms have been developed to aid in discriminating...

The Recording and Research Center, The Denver Center for the Performing Arts, 1245 Champa Street, Denver, CO 80204 Anatomically, the female larynx differs from the male in vocal fold length, thickness, angle of the thyroid laminae, resting angle of the glottis, vertical convergence angle in the glottis, and a number of other ways. The vocal fold tissue itself differs with respect to the thickness of the mucosa and fiber composition, e.g., the ratio of connective tissue to muscle tissue. Biomechanically, it has been found that the stress‐strain curves are slightly more linear in females than males. Based on these and other known differences in morphology and viscoelastic properties, a scaling procedure is attempted for fundamental frequency, intensity, and other acoustic properties. Some rationale is give also for why females develop pathology at different sites in the larynx, given what is currently known about mechanical stress in the tissues.

Vocalis muscle tissue was dissected from live dogs and kept alive in Krebs‐Ringer solution for several hours. The temperature of the tissue was maintained at approximately 37 °C by a circulating fluid around the sample holder. Force‐elongation responses of the vocalis muscle under various levels of stimulated muscle activity were obtained through one‐dimensional stretch and release. Both force and elongation were monitored electronically with a Cambridge Technology Dual Servo System (ergometer), while stimulation was applied to the muscle with fine‐wire electrodes and a Grass stimulator. Families of force‐elongation curves for various frequencies of stimulation and various relaxation times are given, from which average elastic constants, such as Young's modulus, can be estimated. [Work supported by NINCDS.]

A detailed study of normal vibrational patterns of the vocal cords and a preliminary investigation of simple vocal cord pathologies have been conducted. A 16‐mass model described in earlier abstracts [Titze and Strong, J. Acoust. Soc. Amer. 51, 138(A) (1972); 52, 123(A) (1972)] represents mucous membrane, vocal ligament, and vocalis muscle. An 18‐section vocal tract and a 12‐section nasal tract are coupled to this system. Fundamental frequency, vocal dynamics, register, and vibrato are linked quantitatively with the physiological correlates, i.e., subglottal pressure, vocalic contraction, vocal cord length, and elastic properties of vocal ligament and mucous membrane. Interactions between the vocal tract and the glottal source are demonstrated for the vowels /a/ and /i/. An attempt is made to simulate localized mass increase, i.e., a vocal node, and localized stiffening to approximate the effect of a mild cancer. The perturbed vibrational patterns and the corresponding synthesized vowels art, discussed. Samples of normal utterances and slightly pathological utterances are presented for several pitches in modal and falsetto register.

Vocalis muscle and vocal ligament are modeled separately, each by a row of eight masses, with 2 degrees of freedom per mass. This allows for motions in the direction of the flow and transverse to the flow. Vertical phasing at low pitches and horizontal phasing at high pitches are demonstrated. Parameters, which can be adjusted at liberty, are cord elasticity, ligament tension, vocalis tension during contraction, subglottal pressure, and the cord boundary. The variable boundary is employed to allow for various statically strained conditions, such as medial compression and open chink. The normal modes of the system are seen by applying either a sudden burst of pressure (such as in a cough) or by displacing one or several of the masses and allowing the motion to come to rest. Chest, middle, and high registers are demonstrated, including a register transition.

Stress‐strain relationships and Young's modulus were obtained from live excised canine vocal fold tissue under two conditions: immediately following stepwise elongation and after 20 min of stress relaxation. The preparations were oxygenated and maintained in a balanced electrolyte solution at a pH of 7.4 and a temperature of 37°C. Tissue viability was tested via electrical field stimulation. Previously reported studies of the viscoelastic properties of vocal fold muscle have used dead tissue, thus limiting the validity of this information for use in the development of biomechanical models of voice production. Young's modulus is compared with values previously obtained on dead vocal fold tissue [Kakita, Vocal Fold Physiology, edited by Stevens and Hirano (1981)].

The vocal folds may be represented as a viscoelastic system. The power dissipated in the vocal folds (Pf) during phonation results from viscous loss in the vocal fold tissues. It can be approximated on the basis of a simplified representation of the geometry and movement of the fold, and has been estimated to be a significant part of the available pulmonary power [Titze, Tenth Symposium, Care of the Professional Voice (1981)]. Assuming that the energy dissipated is converted into heat, it is added to heat generated by muscle contraction (Pm). Heat is lost by thermal conductivity (Pc) and blood perfusion (Pb). While Pf and Pm increase during phonation, Pb has been shown to decrease during phonation [Hiroto, Otol. Fukuoka 17, 1–5 (1971)]. From estimates of the thermal coefficients of routes by which heat is conducted away from the vocal folds, it is possible to conclude that a temperature gradient should exist between the medial and lateral sections of the folds. Experiments on excised larynges are describe...

A five‐parameter representation of glottal and glottographic waveforms is proposed. Parameters consist of amplitude, period, open quotient, symmetry (speed) quotient, and an exponent of a trigonometric function that governs the sharpness of closure. A compact mathematical formula, based on a nonlinearly distorted sinusoid, is proposed for synthesis. It produces a variety of functions, ranging from idealized square, sawtooth, and sinusoidal waveshapes to more complex truncated and skewed pulses. The effects of each parameter on the waveshapes and spectral envelopes are demonstrated. Sharpness of closure determines the spectral slope, symmetry quotient is related to the amplitude of the spectral ripple, and open quotient is related to the periodicity of the ripple. Strategies for automatic picking of the parameters are suggested for analysis of glottographic waveforms.