Give as many degrees of freedom as possible to a wind player, hook it up to various synthesis algorithms including physical models of wind instruments.
The HIRN wind controller, sensed rotation and translated in both hads, arm orientation, independent control with each finger, breath pressure, and even muscle tension in the lips. Mappings from these controls to the parameters of the WhirlWind meta-wind-instrument physical model allowed exploration of new "spaces" of acoustical process, and the HIRN also was investigated as a controller for FM and other synthesis techniques. Negative lessons from the HIRN project indicated a godd thing, and that attempting to build a "super instrument" with no specific musical composition to directly drive the project yields interesting research questions, but win no real product or future direction. One positive lessons from the project is that the co-design of synthesis/processing algorithms with controllers can benfit both.