Using Driver ' s Speech to Detect Cognitive Workload

In a recent driving simulator study [1], drivers were asked to engage in spontaneous conversations with the remote experimenter. While driving a rather complicated world, they used a hands-free cell phone. Each driver engaged in four “Neutral” and four “Intense” conversations of approximately three minutes each. The objective driving performance and the subjective “workload” showed significant differences between topic types. In the current study, an analysis of the speech during the spontaneous conversations was undertaken to see if there was any correlation between speech patterns and the conversation topic type. This paper discusses the successful results and their implications.