Samples from "DCCRN: Deep Complex Convolution Recurrent Network for Phase-Aware Speech Enhancement
Authors: Yanxin Hu, Yun Liu, Shubo Lv, Mengtao Xing, Shimin Zhang,Yihui Fu, Jian Wu, Bihong Zhang, Lei Xie
Code: https://github.com/huyanxin/DeepComplexCRN
Part1: simulated WSJ0 and MUSAN data
| Model |
0db |
5db |
10db |
15db |
20db |
| Noisy |
|
|
|
|
|
| Clean |
|
|
|
|
|
| LSTM |
|
|
|
|
|
| CRN |
|
|
|
|
|
| DCCRN-R |
|
|
|
|
|
| DCCRN-C |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-CL |
|
|
|
|
|
| DCUNET |
|
|
|
|
|
Part2: DNS test set
Samples on no reverberation set
| Model |
Sample 1 |
Sample 2 |
Sample 3 |
Sample 4 |
Sample 5 |
| Noisy |
|
|
|
|
|
| NSNet |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-E-Aug |
|
|
|
|
|
| DCCRN-CL |
|
|
|
|
|
| DCUNET |
|
|
|
|
|
Samples on reverberation set
| Model |
Sample 1 |
Sample 2 |
Sample 3 |
Sample 4 |
Sample 5 |
| Noisy |
|
|
|
|
|
| NSNet |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-E-Aug |
|
|
|
|
|
| DCCRN-CL |
|
|
|
|
|
| DCUNET |
|
|
|
|
|
Samples on real recording set
| Model |
Sample 1 |
Sample 2 |
Sample 3 |
Sample 4 |
Sample 5 |
| Noisy |
|
|
|
|
|
| NSNet |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-E-Aug |
|
|
|
|
|
| DCCRN-CL |
|
|
|
|
|
| DCUNET |
|
|
|
|
|
Part3: DNS blind set
Samples on no reverberation set
| Model |
Sample 1 |
Sample 2 |
Sample 3 |
Sample 4 |
Sample 5 |
| Noisy |
|
|
|
|
|
| NSNet |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-E-Aug |
|
|
|
|
|
Samples on reverberation set
| Model |
Sample 1 |
Sample 2 |
Sample 3 |
Sample 4 |
Sample 5 |
| Noisy |
|
|
|
|
|
| NSNet |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-E-Aug |
|
|
|
|
|
Samples on real recording set
| Model |
Sample 1 |
Sample 2 |
Sample 3 |
Sample 4 |
Sample 5 |
| Noisy |
|
|
|
|
|
| NSNet |
|
|
|
|
|
| DCCRN-E |
|
|
|
|
|
| DCCRN-E-Aug |
|
|
|
|
|