Search papers, labs, and topics across Lattice.
This paper outlines the DCASE 2026 Challenge Task 2, which focuses on noise-aware unsupervised anomalous sound detection (UASD) for machine condition monitoring, emphasizing the need for reliable detection in noisy environments. The challenge allows participants to utilize two-channel audio samples, capturing sounds from both near and far locations relative to the machine, to enhance the differentiation between machine sounds and environmental noise. Preliminary findings suggest that this dual-channel approach could significantly improve UASD performance in real-world applications where noise is prevalent.
Noise-aware UASD could revolutionize machine monitoring by effectively distinguishing between machine sounds and environmental noise, even in challenging acoustic conditions.
This paper presents an overview of DCASE 2026 Challenge Task 2, titled "Noise-aware unsupervised anomalous sound detection (UASD) for machine condition monitoring." The task aims to advance noise-robust anomalous sound detection for machine condition monitoring under the unsupervised setting, where only normal machine sounds are available for training. Reliable detection under noisy conditions is crucial for practical deployment, but previous DCASE Task 2 settings provided limited information about environmental noise, potentially limiting UASD performance in highly noisy situations. To address this limitation, DCASE 2026 allows participants to exploit two-channel audio samples simultaneously captured at locations near and far from the target machine. Since the distant microphone is expected to contain relatively stronger environmental noise and weaker direct machine sounds, it may help distinguish environmental noise components from the target machine sounds. After the challenge submission deadline, challenge results and an analysis of the submitted systems will be added.
