A biomechanical laryngeal model of voice F0 and glottal width control.

A simplified mathematical model of the larynx, based on biomechanical principles, is described. Components represented include cartilages (cricoid, thyroid, arytenoids, and corniculates), muscles (thyroarytenoid [TA], cricothyroid pars rectus [CTR], cricothyroid pars oblique [CTO], posterior cricoarytenoid [PCA], lateral cricoarytenoid [LCA], and transverse arytenoid [TrA]), ligaments (cricoarytenoid [CAL], anterior cricothyroid [ACTL], posterior cricothyroid [PCTL], and vocal ligaments [VL]), and subglottal pressure (PS). Model outputs included equilibrium positions of cartilages, the glottal width, and the estimated fundamental frequency (F0) of vocal fold vibration. Major findings were that TA, CTR, CTO, and TrA all had substantial effects on F0: that PCA caused glottal opening by rotating the arytenoids laterally; that LCA both ventrolaterally translated and medially rotated the arytenoids, producing minimal effects on glottal closure; and that TrA had major effects on glottal closure by dorsomedially translating and medially rotating the arytenoids. The effects of LCA and PCA were generally diminished as activation of other muscles was increased. Muscle activation plots (MAPs) were used to study the effects of independent parametric variation of several muscles on F0 and glottal width. Both of these parameters were found to be under simultaneous control by TA, CTR, CTO, and TrA. LCA and PCA also had some influence on F0 and glottal width contours, but this appeared to be of limited functional significance, since changes in F0 tended to be offset by changes in glottal closure. Finally, the functional significance of rotation and subduction of the cricothyroid joint was examined. It was found that the combination of subduction with rotation provided greatest control and range of Fo as muscle activation was varied systematically. Strengths and limitations of the current model are discussed, future developments are suggested, and implications of model results as constraints for neural modeling efforts are described.