An Active Machine Hearing System for Auditory Stream Segregation

TL;DR

This paper presents a binaural machine hearing system that performs auditory stream segregation using probabilistic clustering of sound source locations, incorporating head movements to enhance performance in complex auditory scenes.

Contribution

It introduces a novel probabilistic framework based on von Mises distributions for joint localization and segregation, mimicking human auditory stream grouping.

Findings

Effective segregation of multiple sound sources in complex scenes

Improved localization accuracy with head movements

Robust performance with speech and non-speech sounds

Abstract

This study describes a binaural machine hearing system that is capable of performing auditory stream segregation in scenarios where multiple sound sources are present. The process of stream segregation refers to the capability of human listeners to group acoustic signals into sets of distinct auditory streams, corresponding to individual sound sources. The proposed computational framework mimics this ability via a probabilistic clustering scheme for joint localization and segregation. This scheme is based on mixtures of von Mises distributions to model the angular positions of the sound sources surrounding the listener. The distribution parameters are estimated using block-wise processing of auditory cues extracted from binaural signals. Additionally, the proposed system can conduct rotational head movements to improve localization and stream segregation performance. Evaluation of the system is conducted in scenarios containing multiple simultaneously active speech and non-speech sounds placed at different positions relative to the listener.