Synthesis of pathological voices and experiments on the effect of jitter and shimmer in voice quality perception

Simulation of voice disorders has a great deal of interest among clinicians and scientists interested in the simulation of voice timbres. Jitter and shimmer are commonly used as acoustic cues for the analysis and clinical assessment of the voice. In this paper, a voice synthesizer based on direct digital synthesis (DDS) is proposed to model jitter and shimmer. The major advantage of the DDS-based synthesizer compared to other synthesizers is its ability to generate only jitter, shimmer or both. Moreover, these synthetic sounds have a good naturalness. Experiments are carried out in order to compare perceptually voice stimuli with jitter or shimmer only to voice stimuli that include simultaneously jitter and shimmer. The DDS-based synthesizer is used to generate male and female sustained vowels [a] with different amounts of jitter and shimmer. Conducted auditory experiments have shown that stimuli with jitter and shimmer are hoarser than stimuli with jitter or shimmer only.