Self-oscillating Models of the Tongue Tip for Sim- Ulating Alveolar Trills

Alveolar trills are sounds produced by the self-oscillation of the tongue tip in the alveolar region. Specific aeroacoustic conditions around the lingual constriction are required to produce such sounds, but have not been fully investigated dueto the lack of physical modeling. This study discusses the possibility to model the tongue tip behavior with lumped mass-spring systems similar to those used for the vocal folds. It compares a single mass model to a two-mass model with mobile air flow separation point. The tongue tip model is introduced into a transmission line circuit analog model of the vocal tract to which a self-oscillating model of the vocal folds is connected, in order to accurately reproduce the aerodynamic conditions at the vicinity of the lingual constricion. Results show that, although the phase shift between masses with the two-mass model is very small, it differs with the single mass model by accounting for less acoustic interactions with the vocal tract and the vocal folds, and proves to be more robust for the simulation of alveolar trills than the single mass model. The study also proposes to model the incomplete closure of the vocal tract during contacts between the tongue tip and the hard palate in the oscillation process, by connecting an acoustic waveguide both downstream and upstream of the lingual constriction, producing a lateral air leakage. The model presented in the paper is then a useful tool to study the articulatory and physiological conditions required for the production of alveolar trills.