Spherical Array

Spherical arrays are collections of sensors arranged on the surface of a sphere, offering 360-degree coverage for applications like audio-visual localization, sound field synthesis, and 3D scene reconstruction. Current research focuses on developing efficient algorithms for processing data from these arrays, including novel beamforming techniques using real-valued weights and the application of graph-based convolutional neural networks for tasks such as semantic segmentation. This geometry is proving particularly valuable in addressing challenges related to self-motion in egocentric vision and improving the realism and stability of 3D-aware image synthesis, impacting fields ranging from virtual reality to room acoustics modeling.