A physical model for articulatory speech synthesis. theoretical and numerical principles