Nonlinear dynamics of phonations in excised larynx experiments.

Nonlinear dynamic methods including correlation dimension and Lyapunov exponents are applied to quantitatively analyze phonations in excised larynx experiments. Irregular phonations are typically characterized by aperiodic waveforms and broadband spectra. Finite correlation dimensions and positive Lyapunov exponents of irregular phonations demonstrate the existence of chaos in excised larynx phonations. Furthermore, the correlation dimension, maximal Lyapunov exponent, jitter, shimmer, and peak prominence ratio are used to statistically distinguish irregular phonations from normal phonations. The correlation dimension and maximal Lyapunov exponent indicate a significant difference between irregular and normal phonations; however, jitter, shimmer, and peak prominence ratio do not reveal such a significant difference and thus are unsuitable to differentiate between irregular phonations and normal phonations. These findings might potentially assist investigators in understanding rough phonations and developing clinically valuable methodologies for the diagnosis of voice disorders.