A Physical Model for Articulatory Speech Synthesis. Theoretical and Numerical Principles