ITRW on Non-Linear Speech Processing (NOLISP 05) Barcelona, Spain April 19-22, 2005

Synthesis of Disordered Voices

Julien Hanquinet (1), Francis Grenez (1), Jean Schoentgen (2)

(1) Department Signal and Waves, Université libre de Bruxelles, Brussels, Belgium
(2) National Found for Scientific Research, Belgium

This presentation concerns the simulation of disordered voices. The synthesis is based on shaping functions, which are nonlinear memoryless inputoutput characteristics that transform a trigonometric driving function into a synthetic phonatory excitation signal. One advantage of the shaping fuction model is that the instantaneous frequency and the spectral balance of the phonatory excitation signal are controlled by two separate parameters. It is shown how to synthesize different types of dysperiodicities by modulating both the amplitude and instantaneous frequency of the driving function. The voice disorders that are simulated are short- and long-term perturbations of the vocal frequency, biphonation, diplophonia and raucity. Turbulence noise is modeled by additive white noise.

Full Paper

Bibliographic reference.  Hanquinet, Julien / Grenez, Francis / Schoentgen, Jean (2005): "Synthesis of disordered voices", In NOLISP-2005, 168-173.