FILTER is the Freely Improvising, Learning, and Transforming Evolutionary Recombination system. Following in the tradition of Pauline Oliveros’s Expanded Instrument System (EIS) and Doug Van Nort’s Granular-feedback Expanded Instrument System (GREIS), FILTER accepts a fellow performers’ sound and re-presents this material in a transformed state over the course of a performance. FILTER advances this area of research by including machine listening and musical structure-learning capabilities in order to create new musical gestures, phrases, and textures based on what it has learned from its partners’ style. As with any sensitive musician, FILTER can mirror its improvising partner or can suggest new musical directions, relative to the given musical context. The development of FILTER has been supported by grants from the CreativeIT initiative of the National Science Foundation.