A Novel Binaural Cue Preservation Loss for DNN-Based Binaural Speech Enhancement

2026-08-17

AI Summary

This study addresses the issue that deep neural network-based binaural speech enhancement often compromises spatial localization cues. To mitigate this, we propose a novel training objective incorporating binaural reconstruction error and joint cue loss functions. By explicitly constraining inter-channel spectral relationships and jointly modeling interaural level and phase differences (ILD/IPD), this approach optimizes the network to balance noise reduction with spatial fidelity. Experimental results demonstrate that the proposed strategy effectively reduces masking distortion while maintaining robust denoising performance. Notably, it significantly outperforms baseline methods in ILD preservation, achieving an optimal trade-off between noise suppression and accurate retention of spatial auditory cues. Consequently, this work substantially enhances speech enhancement quality in binaural listening scenarios by ensuring precise spatial perception alongside effective interference removal.

Abstract

Binaural speech enhancement for hearing aids aims to reduce noise while preserving the interaural cues needed for spatial localization. Although deep neural network-based methods achieve strong noise reduction, they often distort the rela- tionship between the left and right signals. In this paper, we propose two novel binaural cue preservation losses. First, a binaural reconstruction error loss that directly penalizes masking-induced distortion in the relationship between the left and right spectra, providing a more direct measure of the binaural consistency than conventional separate interaural level differences (ILD) and interaural phase differences (IPD) errors as in prior work. Second, a binaural cue loss that jointly models ILD and IPD to better preserve the binaural structure. Experimental results show that both proposed losses maintain strong noise reduction performance and reduce masking- induced distortion compared to the state-of-the-art baseline cue loss, while the second proposed joint binaural cue loss also outperforms the baseline in ILD preservation.

Problem

• Binaural speech enhancement
• Interaural cue preservation
• Spatial localization
• Deep neural network

Innovation

• Binaural Cue Preservation Loss
• Binaural Reconstruction Error
• Joint ILD-IPD Modeling
• Binaural Speech Enhancement
• Masking-induced Distortion