Voice assistants’ user-friendly response strategies considering information density, response time and response turns

With the development of voice interaction technology, voice assistants (VAs) have been widely used in various smart devices. However, previous studies have paid little attention to how information density affects the VA user experience. Moreover, the interaction effects of information density, response time, and response turns are unclear. Thus, the aim of this study was to investigate the effects of information density, response time and response turns on the VA user experience. Thirty-six participants (18 females and 18 males) were invited for an empirical experiment. Three levels of information density, i.e., low (average information per word less than 13), medium (between 13 and 14), and high (greater than 14), three response times, i.e., short (i.e., 800 ms), medium (i.e., 1200 ms), and long (i.e., 1600 ms), and two response turns, i.e., single-turn and multiturn, were considered. Five measurements were used to evaluate the VA experience, namely, preference for response time, preference for response turns, information quality and relevance, semantic intelligence, and user satisfaction. The results indicated that information density had significant effects on information quality and relevance and user satisfaction. Response time had a significant effect on the preference for response time. The number of response turns had a significant effect on the five measurements. Specifically, the user experience was best for low information density and worst for high information density; a shorter response time led to a better user experience, and users with a single-turn response had a better user experience than those with a multiturn response. Design implications for improving the VA user experience that consider information density, response time and response turns are discussed.