Speech Enhancement on Dataset-1 WSJ0, DEMAND, RNNoise 1.0 (test)
3.44PESQDCUnet-MC
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DCUnet-MCLoss=Proposed Mixed Loss, Attention=CCBAM2021.02 | 3.44 | 97.1 | 19.33 | 23.81 | |
| DCUnet-MLoss=Proposed Mixed Loss2021.02 | 3.38 | 96.85 | 18.59 | 22.1 | |
| DCCRN-MCLoss=Proposed Mixed Loss, Attention=CCBAM2021.02 | 3.34 | 96.76 | 18.42 | 24.18 | |
| DCCRN-MLoss=Proposed Mixed Loss2021.02 | 3.28 | 96.3 | 18.03 | 23.06 | |
| DCUnetArchitecture=DCUnet-202021.02 | 3.25 | 96.26 | 18.31 | 20.16 | |
| DCCRNFramework=Real-time CRN, Complex-valued=true2021.02 | 3.17 | 95.8 | 17.71 | 21.35 | |
| NoisyProcessing=None2021.02 | 1.97 | 87.83 | 6.28 | 10.64 |