Abstract:

In this pilot study, the authors employed a basic driving simulator to examine both driving behavior and task performance as subjects performed music retrieval tasks using one of three variants of the "SpeakPod" voice search prototype. The variants shared speech and manual interface designs but differed in visual output capabilities. Preliminary data indicate that the chosen variant and the presence or absence of a music search task had little impact on the chosen driving metric. Post-drive NASA-TLX survey results do not show any of the three variants to be any more cognitively demanding than any other. There was also no clear winner in terms of task success rate.