How Task Complexity and Voice Type Shape Prosodic and Physiologic Cues of Engagement in Human-Machine Dialogue

Overview & Synopsis

This study investigates prosodic cues of user engagement in task-oriented interactions with non embodied conversational assistants. We hypothesize that task complexity and the type of assistant voice (synthetic vs. human) influence user engagement. We measure both vocal and physiological parameters expected to correlate with engagement. We focus on prosodic features such as fundamental frequency, speech rate and intensity, and also explore physiological indicators, including electrodermal activity and heart rate. While we did not observe significant variations in physiological parameters, our results suggest that both voice type and task complexity influence prosodic markers of engagement.