Tongue bracing for improved control of vocal-tract model

We have developed a series of physical models of the human vocal tract and sound sources for different purposes, including education, speech and language pathology, scientific studies, and many other applications. Some of the models have dynamic aspects of speech production and can simulate articulatory movements. When the tongue of a dynamic model approaches the palate, tongue bracing helps to form a midline groove on the surface of the tongue, especially for high vowels. Recently, we have fine-tuned the design of our version of Umeda and Teranishi’s (UT) model to simulate continuous speech production. For the series of UT models, a set of plates or blocks line up next to each other to simulate the lower lip and tongue. The movements of these plates/blocks simulate the articulatory gestures. In the new model, the lateral and midline portions of each block for the tongue can be controlled independently. This allows us to achieve tongue bracing and eventually minimizes errors between the target and the actual positions of the blocks with quick motions.