ICASSP 2026 URGENT Speech Enhancement Challenge
Chenda Li, Wei Wang, Marvin Sach, Wangyou Zhang, Kohei Saijo, Samuele Cornell, Yihui Fu, Zhaoheng Ni, Tim Fingscheidt, Shinji Watanabe, Yanmin Qian
TL;DR
The paper tackles the problem of universal speech enhancement that remains effective across diverse distortions, languages, and input formats. It presents the ICASSP 2026 URGENT Challenge with two tracks: Track 1 for universal SE and Track 2 for MOS-based speech quality assessment, emphasizing data curation and speech diversity. The evaluation protocol combines objective metrics, subjective MOS/CMOS testing, and Friedman-style significance, supported by large-scale, curated datasets and strong baselines such as BSRNN, FlowSE, and Uni-VERSA-Ext. Results indicate that hybrid generative–discriminative architectures and high-quality, curated training data yield the best generalization across distortions and languages, advancing both universal SE and perceptual quality assessment toward real-world applicability.
Abstract
The ICASSP 2026 URGENT Challenge advances the series by focusing on universal speech enhancement (SE) systems that handle diverse distortions, domains, and input conditions. This overview paper details the challenge's motivation, task definitions, datasets, baseline systems, evaluation protocols, and results. The challenge is divided into two complementary tracks. Track 1 focuses on universal speech enhancement, while Track 2 introduces speech quality assessment for enhanced speech. The challenge attracted over 80 team registrations, with 29 submitting valid entries, demonstrating significant community interest in robust SE technologies.