ABSTRACT
Automotive speech interfaces are an integral part of a driver's multimodal interface. The interface poses a unique technological and design challenge because of the automotive context. In this chapter, we describe an integrative approach to its design and productization. While our approach is based on User Centered Design, it combines with it technological innovation and theoretical research in areas related to the user, the machine, and the user interaction. Current development trends in speech applications, such as an increase in the number of features and tasks desired by users and the availability of new technologies lead to our vision: Mixed-initiative natural language speech interfaces as part of a multimodal interface allowing the user to combine or switch among the manual-visual and auditory modalities seamlessly. We suggest four areas in need of research: user and context modeling, global markets and cultural variation, dialog management, and multimodal speech interfaces. We introduce six research projects, five of which were performed in collaboration with General Motors and one by Nuance and IBM.
