The ICASSP SP Cadenza Challenge: Music Demixing/Remixing for Hearing Aids

TL;DR

This paper presents the design, results, and insights from the 2024 ICASSP SP Cadenza Challenge, which focused on music demixing and remixing techniques to improve audio quality for hearing aid users in stereo playback scenarios.

Contribution

It introduces a new challenge for music demixing/remixing tailored for hearing aids and analyzes various approaches, highlighting the effectiveness of ensemble models and fine-tuning pretrained systems.

Findings

9 systems outperformed the baseline.

Ensemble of models achieved the best results.

Causal approaches performed worse than non-causal ones.

Abstract

This paper reports on the design and results of the 2024 ICASSP SP Cadenza Challenge: Music Demixing/Remixing for Hearing Aids. The Cadenza project is working to enhance the audio quality of music for those with a hearing loss. The scenario for the challenge was listening to stereo reproduction over loudspeakers via hearing aids. The task was to: decompose pop/rock music into vocal, drums, bass and other (VDBO); rebalance the different tracks with specified gains and then remixing back to stereo. End-to-end approaches were also accepted. 17 systems were submitted by 11 teams. Causal systems performed poorer than non-causal approaches. 9 systems beat the baseline. A common approach was to fine-tuning pretrained demixing models. The best approach used an ensemble of models.