Understanding Everyday Communication Patterns, Stress, and Productivity in Hybrid Work through Egocentric Wearable Audio Sensing

Abstract

The hybrid work trend in the post-COVID period has transformed how people communicate while at work. However, there is a lack of quantitative understanding of how home and office environments shape communicative behavior and affective well-being. To study this, as a part of a larger study of the role of work environment on stress, we designed an egocentric wearable audio sensing system that captured interpersonal communication across both working locations. The system extracts real-time acoustic features on a wearable audio sensor and performs offline foreground speaker diarization to derive speaking load and activeness features. To improve diarization performance under low-dimensional feature constraints, we apply a teacher-student knowledge distillation framework that transfers knowledge from large-scale speech foundation models into a compact ResNet-based model, achieving consistent improvements in diarization performance over the baseline across three public egocentric speech datasets. In a four-month longitudinal deployment with 16 participants, our system revealed location-dependent differences in work communication patterns. Moreover, the mixed linear model analyses showed that office-based speaking load was significantly associated with higher self-reported stress and lower self-reported productivity, while speaking activeness during home-based work amplified both productivity and affective load in this cohort.