This paper discusses the Birl, an electronic wind instrument developed by the authors. It uses artificial neural nets to apply machine learning to the mapping of fingering systems and embouchure position. The design features of the instrument are described, and the machine learning mapping strategy is discussed.