Leveraging LLM and Text-Queried Separation for Noise-Robust Sound Event Detection

Benchmark Model Rank Results
sound-event-detection-on-wilddesedCRNN (with BEATs + Separation)#1PSDS1 (-5dB): 0.134PSDS1 (0dB): 0.219PSDS1 (5dB): 0.291
sound-event-detection-on-wilddesedCRNN (with BEATs)#2PSDS1 (-5dB): 0.065PSDS1 (0dB): 0.138PSDS1 (5dB): 0.236