User’s Needs for Voice Agents in a Smart Home Context

Abstract

Background As the smart home market expands, there has been substantial research conducted on technology and business, but insufficient research on the user’s perspective. Therefore, this research was aimed to uncover the user’s needs of voice agents in the context of smart home. We analyzed the interventions of the voice agents in terms of their roles and forms. The goal of this study was to derive design implications that can be considered when devising smart home service with voice agents. Methods To understand the user's needs in a smart home environment, we conducted experiments in which users spoke and requested commands to the voice agents. We used methodologies of cultural probes and in-depth interviews. Participants who were unable to visit their homes were provided with pictures of the residential environment so that they could be immersed in the experiment. Through the experiments, we gained interview and speech data from the participants. Interview data were analyzed in an inductive approach, and speech data in a deductive approach. We confirmed the validity of these data based on the triangulation method. Results There were 32 core needs required in a smart home environment, and the roles users expected from voice agents were divided into two categories: ‘Analysis and Management of Residential Environment’ and ‘Daily Care and Support’. It was also derived that the form of the voice agent intervention needed to operate differently depending on each core need, such as active and passive voice. In addition, based on speech data, we categorized the characteristics of smart home services from the user's perspective, and calculated the percentage of speech per residential space. Conclusions This research proposed the implications that should be considered when designing voice agents for smart home services, such as ‘the form of voice agent’s intervention based on user’s needs’, ‘the data input method and structure’, and ‘the necessity of design considering multi-user and personalization’.