Proceedings of Meetings on Acoustics

This paper presents recent progress in research examining the mechanisms of voice generation and a method for physiologically-based speech synthesis. The overarching goal of this research is the precise modeling of interactions among physical systems involved in the processes underlying voice generation. In the basic voice generation system, a flow-structure interaction between glottal flow and the vocal folds causes self-oscillations of the folds, where flow separation, a nonlinear aerodynamic phenomenon, plays an important role. The fluid dynamic theory implies that a thin boundary layer formed near the glottal wall characterizes the flow behavior, including flow separation, jet formation, and pressure loss across the channel. We therefore use the interactive boundary layer method to analyze glottal flow and show how the flowstructure interaction is effective in maintaining vocal fold oscillations. In addition, the interaction between the voice generation system and the vocal-tract filter, i.e., the source-filter coupling, has been found to involve nonlinear factors in speech, such as skewing of glottal flow, unsteadiness in vocal fold oscillations, and transitions in voice register.