Takuya Fujimura, Yoshiki Masuyama, Gordon Wichern, Christoph Boeddeker, Julius Richter, Jonathan Le Roux, "NABEATs: Noise-Aware Audio Representation Learning" https://t.co/DIKPsx2h3K
Our team at @merl_news ranked 1st🥇 in Task 2 and 3rd🥉 in Task 4 at the DCASE 2026 Challenge: https://t.co/wZPfbpw4pi
Task 2 submission was led by our intern @i15fujimura1t, while @kohei_1979 mainly worked on Task 4.
We also achieve the best audio tagging accuracy in Task 4🎉