LAUSR

Speech production involves air traveling from a reservoir (the lungs) through various ducts (the bronchi and trachea, the vocal and nasal tracts), past various structures at which sounds can be generated by the flow of air. Some structures (e.g., the vocal folds or the tongue tip) can be positioned so as to self-oscillate, generating quasi-periodic sound; some (e.g., the lips or the tongue tip) can be positioned to create a constriction small enough to generate turbulence, and thus turbulence noise. These sounds excite the resonances of the tract. They can also interact, as when the sound generated by vocal fold vibration modulates the turbulent jet formed at a downstream constriction. It would seem to be straightforward to predict the sound that propagates through the tract and is radiated externally, except that the tract changes shape continually; the walls of the tract vary in compliance, from hard (e.g., teeth) to yielding (e.g., cheeks); and in spite of co-opting many types of medical imaging, it is very difficult to measure the interior shape of the tract accurately and without disturbing either the speech or the speaker. Some acousto-fluidic interactions that occur during speech, and the ways they have been modeled, are discussed.