Self-supervised Neural Audio-Visual Sound Source Localization via Probabilistic Spatial Modeling
This addresses the challenge for autonomous robots to understand environments by distinguishing sound sources with similar appearances, though it is incremental as it builds on prior self-supervised audio-visual methods by incorporating spatial audio.
The paper tackled the problem of localizing sound source objects in visual scenes without manual labeling by proposing a self-supervised method using 360° images and multichannel audio signals, and it demonstrated localization of multiple speakers in simulated data and detection of objects like talking visitors in real museum data.
Detecting sound source objects within visual observation is important for autonomous robots to comprehend surrounding environments. Since sounding objects have a large variety with different appearances in our living environments, labeling all sounding objects is impossible in practice. This calls for self-supervised learning which does not require manual labeling. Most of conventional self-supervised learning uses monaural audio signals and images and cannot distinguish sound source objects having similar appearances due to poor spatial information in audio signals. To solve this problem, this paper presents a self-supervised training method using 360° images and multichannel audio signals. By incorporating with the spatial information in multichannel audio signals, our method trains deep neural networks (DNNs) to distinguish multiple sound source objects. Our system for localizing sound source objects in the image is composed of audio and visual DNNs. The visual DNN is trained to localize sound source candidates within an input image. The audio DNN verifies whether each candidate actually produces sound or not. These DNNs are jointly trained in a self-supervised manner based on a probabilistic spatial audio model. Experimental results with simulated data showed that the DNNs trained by our method localized multiple speakers. We also demonstrate that the visual DNN detected objects including talking visitors and specific exhibits from real data recorded in a science museum.