ABSTRACT
Automatic speech recognition (ASR) can potentially improve the usability and safety of operating in-car systems such as phones, entertainment systems and intelligent transportation systems (ITS). However, its success relies on the careful design of the interface to match the abilities, expectations and preferences of the user population. This paper reviews some of the human factors issues in vocabulary and dialogue design, and considers the interfaces to some real applications. A study is then reported in which drivers were consulted as to their preferred interactions with in-car ASR systems. A main finding was that users preferred to issue terse, goal-directed commands rather than engage in natural language dialogues.
