Design of Superior Systems and Research Programs Needed to Achieve Them

Chairs: Pierre Divenyi, Nat Durlach, Dan Ellis, Deliang Wang

After having listened, commented, argued, or just patiently sat at the eleven preceding sessions, we all must have at least some ideas as regards solving the one major problem: How to build a machine that would, on its own or in conjunction with an active listener, comprehend (or help comprehend) speech by separating it from the complex acoustic environment in which it appears. We hope to come up with a few scenarios and discuss each of them in sufficient depth to arrive at a mutual understanding of the direction in which research striving to achieve the main objective should proceed, if given the necessary encouragement by various funding sources.