Information Sci. Res. Lab., NTT Basic Res. Labs., 3-9-11 Midori-cho, Musashino-shi, Tokyo 180, Japan
The generation of articulatory movements suffers from the degrees-of-freedom problem in determining the phoneme-related shape of the vocal tract. This model is intended to study the coordinative movement of articulatory organs. The redundancies of the articulation system are solved by minimizing an appropriate objective function to determine articulatory movement uniquely. In the model, articulatory movements are represented as the output of a multi-dimensional second-order linear system driven by input forces. These movements are partially constrained by phoneme-related features of the vocal tract shape at given instants. The input forces are determined by minimizing the cost function, represented as the energy sum of the time-differentiated system inputs and outputs. Then, articulatory movements are obtained as the system response to the optimum inputs. The trajectory formation model is computer-simulated to generate tongue, lip, and jaw movements for VCV segments. Comparison of simulated trajectories with data measured using a magnetic position-sensing device shows that the model is able to generate accurate jaw--lip or jaw--tongue coproductions and anticipatory coarticulation for the segments.