BiEAR: A Human Auditory-Inspired Adaptive Binaural Front-end for Multi-Speaker Localisation and Distance Estimation

2026-06-04

AI Summary

This study addresses the limited robustness of multi-speaker direction-of-arrival and distance estimation in complex acoustic scenes under unknown speaker identities and room environments. Inspired by the medial olivocochlear efferent feedback mechanism in the human auditory system, this work proposes a neural-controlled adaptive binaural auditory front-end model. It is the first to incorporate a human-like adaptive mechanism into binaural processing, dynamically adjusting the frequency selectivity of filter banks to generate time–frequency adaptive representations during inference, thereby emphasizing information-rich spectral bands. Experimental results demonstrate that the proposed method significantly outperforms fixed front-end approaches in both anechoic and real-room conditions, achieving notable improvements in multi-speaker localization accuracy and generalization capability to unseen acoustic environments.

Abstract

We present BiEAR, a human auditory-inspired adaptive binaural front-end for multi-speaker localisation and distance estimation. Inspired by medial olivocochlear (MOC) feedback in human hearing, BiEAR uses a neural controller to adaptively adjust the frequency selectivity of a binaural auditory filterbank during inference. This yields time-frequency adaptive representations for ears, enabling the model to respond to changing acoustic conditions. We evaluate BiEAR on multi-speaker localisation and distance estimation in anechoic and real-room environments. Results show that the adaptive front-end improves localisation accuracy and robustness to unseen speakers and rooms compared with commonly used fixed binaural front-ends. Visualisation and analysis of learned filter adaptations show that BiEAR emphasises informative frequency bands over time. These findings suggest that adaptive, biologically inspired binaural front-ends can improve machine hearing robustness in complex acoustic scenes.

Problem

• multi-speaker localisation
• distance estimation
• binaural front-end
• acoustic robustness
• complex acoustic scenes

Innovation

• adaptive binaural front-end
• medial olivocochlear feedback
• multi-speaker localisation
• distance estimation
• biologically inspired hearing