A comparison of voice quality measures across in-ear, outer-ear, and standard microphone recordings
Xinyi Zhang, Alessandro Braga, Arian Shamei, Rachel E. Bouserhal · The Journal of the Acoustical Society of America · 2024
Voice quality provides significant insights into one’s health. Recent advancements in intra-aural devices have enabled the longitudinal monitoring of speech and its changes. These wearables can record speech from in-ear microphones (IEMs) and outer-ear microphones (OEMs). The laboratory gold standard of speech recording uses a high-sensitivity low-noise microphone placed in front of the mouth to capture signals accurately. Speech recorded inside an occluded ear canal differs considerably from this due to the bone-and-tissue conduction and the effect of ear occlusion. OEMs record speech transmitted solely through air like the standard microphones but are more influenced by indirect air conduction due to their positioning. This study compares voice quality measures across IEM, OEM, and the standard microphone (REF) using an open-access database. We employed linear mixed-effect modeling to analyze the effects of different microphone recordings on these metrics. Results indicate that while pitch control remains relatively intact, IEM and OEM exhibit varying deviations from REF. Overall, OEM resembles REF in jitter, whereas IEM is higher. IEM resembles REF in shimmer, while OEM is lower. For HNR, both IEM and OEM are higher than REF. Sex-based difference was also observed, and the correlation between microphone differences and fundamental frequency was explored.