Ego Noise

Ego-noise, encompassing sounds generated by a system itself (e.g., robot motors, UAV propellers), is a significant challenge in various applications requiring audio processing and perception. Current research focuses on developing methods to reduce or utilize ego-noise, employing techniques like multichannel Wiener filtering, variational autoencoders, and latent diffusion models, often in conjunction with environmental noise reduction. These advancements are crucial for improving speech enhancement in robotics and autonomous systems, enabling more robust human-robot interaction and potentially facilitating novel sensing capabilities like acoustic reflector detection for collision avoidance.