Developing the Proactive Speaker Prototype Based on Google Home

Abstract

Smart speakers and conversational interfaces increasingly make it into consumer’s homes. Listening to users’ commands assigns them a rather passive role. Proactive speakers, on the other hand, have the potential to empower a broad range of applications such as context-aware health interventions and self-tracking systems. To achieve proactivity, however, requires the speaker to become context-aware and be able to detect opportune moments to initiate interactions. In this work, we present a proactive speaker prototype based on Google Home to investigate interaction contexts. We, additionally, built a voice-based Experience Sampling Method (ESM) application to study contextual factors that are correlated to opportune moments for device-triggered interactions. Preliminary results from a three-week field study with 7 participants indicate the proposed prototype is a robust way to achieve proactivity and implement voice-based ESM. We present use cases for future research and applications for proactive smart speakers in the context of digital health.