C-SL: Contrastive Sound Localization with Inertial-Acoustic Sensors
Human brain employs perceptual information about the head and eye movements to update the spatial relationship between the individual and the surrounding environment. Based on this cognitive process known as spatial updating, we introduce contrastive sound localization (C-SL) with mobile inertial-acoustic sensor arrays of arbitrary geometry. C-SL uses unlabeled multi-channel audio recordings and inertial measurement unit (IMU) readings collected during free rotational movements of the array to learn mappings from acoustical measurements to an array-centered direction-of-arrival (DOA) in a self-supervised manner. Contrary to conventional DOA estimation methods that require the knowledge of either the array geometry or source locations in the calibration stage, C-SL is agnostic to both, and can be trained on data collected in minimally constrained settings. To achieve this capability, our proposed method utilizes a customized contrastive loss measuring the spatial contrast between source locations predicted for disjoint segments of the input to jointly update estimated DOAs and the acoustic-spatial mapping in linear time. We provide quantitative and qualitative evaluations of C-SL comparing its performance with baseline DOA estimation methods in a wide range of conditions. We believe the relaxed calibration process offered by C-SL paves the way toward truly personalized augmented hearing applications.