Experimental research on human phonation is negatively affected by the complexity of the process and the limiting anatomy of the larynx. Numerical simulation of the vocal folds and the formulation of an inverse problem is one way to remedy this. Several studies have explored this choosing different degrees of freedom (DOFs) in a biomechanical Two-Mass-Model (2MM). The selection of the DOFs in an inverse problem has a critical impact on the quality of possible solutions, but also affects the complexity of the problem and convergence speed in solving it. This work compares previous DOF configurations with several new extended configurations in solving the inverse problem for vocal fold recordings of 20 healthy female subjects. Results indicate that, for the 2MM, uncoupled mass and stiffness, and variable collision strength and damping coefficients improve the matching capabilities. They match physiology and lead to up to 50% smaller errors even for a low number of model evaluations.
CITATION STYLE
Gómez, P., Kniesburges, S., Schützenberger, A., Bohr, C., & Döllinger, M. (2017). Degrees of freedom in a vocal fold inverse problem. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10208 LNCS, pp. 475–484). Springer Verlag. https://doi.org/10.1007/978-3-319-56148-6_42
Mendeley helps you to discover research relevant for your work.