A Taxonomy of Noise in Voice Self-reports while Running

Abstract

Smart earables offer great opportunities for conducting ubiquitous computing research. This paper shares its reflection on collecting self-reports from runners using the microphone on the smart eSense earbud device. Despite the advantages of the eSense in allowing researchers to collect continuous voice self-reports anytime anywhere, it also captured noise signals from various sources and created challenges in data processing and analysis. The paper presents an initial taxonomy of noise in runners' voice self-reports data via eSense. This is based on a qualitative analysis of voice recordings based on eSense's microphone with 11 runners across 14 in-the-wild running sessions. The paper discusses the details and characteristics of the observed noise, the challenges in achieving good-quality self-reports, and opportunities for extracting useful contextual information. The paper further suggests a noise-categorization API for the eSense or other similar platforms, not only for the purpose of noise-cancellation but also incorporating the mining of contextual information.