Spacing Out: On the Reliability of Binaural Music Source Separation Metrics

2026-07-28

AI Summary

This study addresses the degradation of spatial audio quality in existing binaural music source separation methods and the lack of objective spatial distortion metrics aligned with human auditory perception. Through perceptual experiments comparing binaural and stereo separation outputs, the work systematically evaluates the effectiveness of prevailing spatial distortion measures and provides an in-depth analysis of various interaural time difference (ITD) estimation algorithms under noise and separation artifacts. The research reveals, for the first time, a significant discrepancy between current objective metrics and subjective listener judgments, particularly highlighting a trade-off between robustness and accuracy in ITD estimation for narrowband instruments such as bass. These findings offer crucial insights for developing high-fidelity, perceptually grounded, and interpretable spatial evaluation frameworks tailored to binaural music processing.

Abstract

Despite the rising popularity of immersive audio, binaural music remains underexplored in music information retrieval (MIR), particularly regarding the task of music source separation (MSS). While existing stereo MSS models can process binaural audio, they often degrade the spatial quality of the separated stems and undermine listener immersion. Through a perceptual study comparing binaural and stereo MSS outputs, we evaluate how well objective spatial distortion metrics correlate with human perception. Our findings reveal varied agreement between these metrics and human judgment, highlighting a lack of reliability when used to evaluate binaural music tasks. Specifically, we find that Interaural Time Difference (ITD) estimation is highly sensitive to noise and separation artifacts. In evaluating two alternative ITD estimation methods, we uncover a critical trade-off between robustness and accuracy, particularly for narrow-band instruments like bass. These results underscore the need for accurate, interpretable spatial metrics designed for binaural music to develop models that preserve source localization and listener immersion.

Problem

• binaural music
• music source separation
• spatial metrics
• Interaural Time Difference
• immersive audio

Innovation

• binaural music source separation
• spatial metrics
• Interaural Time Difference (ITD)
• perceptual evaluation
• immersive audio