Vowel-vowel Planning in Acoustic and Muscle Space

The objective of this research is to investigate the planning and control of vowel-to-vowel sequences using a computer model of the vocal tract. The vocal tract model, which is based on a French male speaker, consists of a biomechanical tongue model (Perrier et al., 2003) with the addition of jaw (rotation and translation) and lip (opening and protrusion) movements. A comparison of acoustic data from the subject on whom the model is based to simulation results from the model for /i/-/a/, /i/-/e/, and /i/-/u/ movements is presented. Results show that planning a V-V sequence in either acoustic space (F1, F2, and F3) or motor space (in terms of lambda values according to the equilibrium point hypothesis, EPH) produces formant trajectories similar to those of the subject's data, and at the same time has a smooth progression of muscle lengths between the two vowels. However, the modeling does not account for all aspects of the data.