The goal of this project is to develop voice input/speech output interfaces for embedded system applications. Our development efforts will be specifically focused on the domain of human interaction with search engines for the World-Wide-Web. Our research is focused in two areas. First, we are investigating issues relating to discourse management including focus of attention, wake-up/sleep mechanisms, error detection/recovery, and contention resolution.  Second, we are attempting to exploit the domain knowledge that is incorporated into contemporary search engines to design simple discourses directed at refining a search.