Sound Event Separation on FSD-Kaggle 2 Sound 2018
14.32SI-SDRTF-Locoformer
Evaluation results
| Method | Links | |
|---|---|---|
| TF-LocoformerLearning Paradigm=Supervised2025.12 | 14.32 | |
| Conv-TasNetLearning Paradigm=Supervised2025.12 | 11.73 | |
| UnasDiffLearning Paradigm=Unsupervised, Strategy=Gradient+Initialization2025.12 | 10.44 | |
| DSG’s σ(t)-prop. γ(t)Learning Paradigm=Unsupervised, Strategy=Gradient+Initialization2025.12 | 9.48 | |
| DPS’s constant γ(t)Learning Paradigm=Unsupervised, Strategy=Gradient+Initialization2025.12 | 4.63 | |
| Analytical SamplingLearning Paradigm=Unsupervised, Strategy=Analytical Sampling2025.12 | 1.01 | |
| Unprocessed2025.12 | -0.01 |