Date of Award

Summer 8-12-2022

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering


Xudong Zheng

Second Committee Member

Qian Xue

Third Committee Member

Yingchao Yang

Additional Committee Members

Liping Yu

Jian-Xun Wang


This study aims to develop knowledge about the roles of intrinsic laryngeal muscles on voice control in both healthy and disordered conditions through comprehensive computational models. The phonation simulator was built by combining a three-dimensional high-fidelity MRI-based model of the larynx, active muscle mechanics, and fluid-structure-acoustic interaction model, which enabled the exploration of the underlayer mechanisms of the link between individual and/or group muscles contractions under both symmetric and asymmetric activations, vocal fold posture, vocal fold vibration, and voice outcomes during voice production.

The first part of this research extensively investigated the effects of cricothyroid and thyroarytenoid muscle activations on voice characteristics through a parametric study. The role of these intrinsic muscles in the adjustment of geometrical and mechanical properties of vocal fold pre-phonatory posture, glottic flow aerodynamics, and acoustic and how all these components

interact were explored. Results were comprehensively validated, and the link between elements of phonation was described in detail.

In the next step, due to the model's ability in the individual muscle activations, unilateral vocal fold paralysis was simulated, and the characteristics of disordered voice were analyzed. The voice simulator was then combined with the implant insertion model and genetic algorithm method to build a computational framework for patient-specific surgical planning of type 1 thyroplasty. This surgery is a standard procedure for treating unilateral vocal fold paralysis; however, it is subject to challenges mainly due to the small size of the implant and the high sensitivity of the voice outcome to the implant shape and position. Therefore, although the patient's voice could be improved, the results might not be as satisfying as expected. Despite actual surgery, with very little room for try and error, the ideal implant could be achieved by optimizing the implant based on the patient's desired voice using the presented computational framework. Both healthy and diseased cases and the corrected case using the optimized implant were simulated. Results revealed that the optimized implant could restore the aerodynamic and acoustic features of the disordered voice in producing a sustained vowel utterance. Furthermore, the performance of the implant in the pitch gliding test, which was simulated using temporal activation of the cricothyroid and thyroarytenoid muscles based on the first part of the study, was evaluated.

In the final step, a physics-informed neural network-based algorithm was presented to reconstruct the three-dimensional cyclic vibration of vocal fold using two-dimensional sparse experimental data and laws of physics. Key acoustic parameters and vibratory dynamics of vocal folds and other parameters, such as flow rate, pressure distribution, and contact force, which are difficult to measure experimentally, were successfully predicted.

Files over 10MB may be slow to open. For best results, right-click and select "save as..."