Using Kansei Design to Create a Predictive Model for Voice User Interfaces for Electronic Appliances

Abstract

Nowadays, virtual voice assistants, are present in various devices – from smartphones to smart speakers and, soon, in all electronic appliances. In this context, it is a significant increase in the user experience with them to make them more engaging. Virtual assistants usually have the same voice tone to give users information. Thus, companies are not fully taking advantage of voice properties and the power they hold on to communication, increasing the user experience and the market product success. In this study, we used the Kansei Design method to predict the emotional user reaction impact of different voice user interfaces for electronic appliances. Resorting to some literature review and online post-research, we defined voice characteristics to manipulate – gender, cadence, and inflection. Related to the semantic space (Kansei words) of the user's perceptions with virtual assistant voices – pleasure, proximity, and arousal. Eighty-three participants (67,5% female and 32,5% male) answered an online questionnaire with 12 possible combinations between gender; cadence; and inflection, with the Kansei words: pleasure; proximity; and arousal. Results vary with the semantic space, but they suggest that people felt more comfortable and relaxed hearing a male voice than a female one. Regarding cadence, a typical speech flow was where people felt more intimate with the voice. Though, participants felt more activated while hearing a female voice, speaking at a higher speed and with inflection in her voice. We generated some use cases with these results to understand how they can guide design processes regarding voice-user interfaces.

Keywords: Voice Assistance, Smart devices, Kansei Design, Predictive models

DOI: 10.54941/ahfe1003372