This chapter reports the use of speech as an input and output medium, in the General Intelligent Driver Support (GIDS) project of the DRIVE programme. One of the aims of GIDS is to identify techniques for supporting driver interaction with a range of in-car electronic devices. A driver's hearing and speech may be regarded as suitable candidates for use in tasks performed during the primary driving task. Special techniques are needed to handle the drawbacks of speech output in a way convenient to drivers. The problems here include background noise and the effects of stress and cognitive loads on recognition. User interfaces to in-car devices have a combination of user-initiated and system-initiated interaction. Some experimental studies were performed, to address specific issues in the design of user interfaces for in-car use. Two of these studies are described: (1) detecting and adapting to stress; (2) encoding message meaning through phrasing and intonation. Guidelines for in-car user interface design include: (1) do not force interaction; (2) support interruption and recovery; (3) provide context markings for interactions; and (4) balance consistency with adaptivity. A sample dialogue is included, that illustrates some of these design features. Outstanding issues of this research include cultural adaptation and long-term evaluation.
Abstract