Towards the simulation of pathological voice qualities

Abstract

The presentation concerns a synthesizer for disordered voices. The simulation of dysphonia is a topic the relevance of which is growing, but to which few studies have been devoted. The simulator that is discussed here involves a nonlinear memory-less model of the glottal area that is driven by a harmonic excitation the instantaneous frequency and amplitude of which are controlled. The glottal airflow rate is generated by means of an aerodynamic model of the glottis, which also comprises trachea-source and source-tract interactions as well as the generation of turbulence noise at the glottis. Trachea and vocal tract are modelled by means of a concatenation of lossy cylindrical pipes of identical length, but different cross-sections. The text concerns the presentation of the synthesizer and its synthetic output, as well as the results of a perceptual evaluation of the naturalness of simulated speech sounds in the framework of a stimuli comparison paradigm.