Artisynth: an extensible, cross-platform 3d articulatory speech synthesizer

We describe our progress on the construction of a combined 3D face and vocal tract simulator for articulatory speech synthesis called ArtiSynth. The architecture provides six main modules: (1) a simulator engine and synthesis framework, (2) a two and three-dimensional model development component, (3) a numerics engine, (4) a graphical renderer, (5) an audio synthesis engine and (6) a graphical user interface (GUI). We have created infrastructure for creating vocal tract models based on combinations of rigid body, spring-mass, and finite element models, and some parametric models. Our infrastructure provides mechanisms to “glue” these and other model types together to create hybrids. Dynamical models whose equations of motion are integrated numerically and animatable parametric models are combined in a single framework. Using ArtiSynth we have created a complex, dynamic jaw model based on muscle models, a parametric tongue model, a face model, two lip models, and a source-filter based acoustic model linked to the vocal tract model via an airway model. These have been connected together to form a complete vocal tract that produces speech and is drivable both by data and by dynamics.