A physically informed glottis model for real glottal flow wave form reproduction

Abstract

A physically informed model of the glottal source is proposed. The model relies on a lumped mechanoaerodynamic scheme based on the mass-spring paradigm. The vocal folds are represented by a mechanical resonator plus a delay line which takes into account the vertical phase differences. The vocal fold displacement is coupled to the glottal flow by means of a nonlinear subsystem, based on a general parametric nonlinear model. The principal characteristics of the flow-induced oscillations are retained, and the overall model is suited for an identification approach where real (inverse filtered) glottal flow signals are to be reproduced. A data-driven identification procedure is outlined, where the parameters of the model are tuned in order to accurately match the target wave form. A nonlinear regression algorithm is used to train the nonlinear part. A set of inverse-filtered glottal flow wave forms with different pitch, open phase/closed phase ratio, and shape of the glottal wave form period, are used to test the effectiveness of the approach. The results demonstrate that the model can reproduce a wide range of target wave forms. Moreover, the flow wave forms generated by the trained model are characterized by spectral richness and are perceived as natural. [Work supported by EU.]