Table 3 Performance of the pre-trained model and the fine-tuned models with ER recorded echo
From: Nonlinear residual echo suppression based on dual-stream DPRNN
| Echo | Model | PESQ | SDR | STOI |
|---|---|---|---|---|
| Artificial speech | LAEC | 1.48 | −2.60 | 0.622 |
| Time | 2.61 | 12.3 | 0.866 | |
| Time_1 | 2.56 | 12.2 | 0.865 | |
| Time_2 | 2.57 | 12.1 | 0.864 | |
| TF | 2.75 | 12.4 | 0.880 | |
| TF_1 | 2.70 | 12.4 | 0.875 | |
| TF_2 | 2.69 | 12.3 | 0.875 | |
| Artificial music | LAEC | 1.48 | −2.90 | 0.634 |
| Time | 2.50 | 11.5 | 0.842 | |
| Time_1 | 2.44 | 11.4 | 0.841 | |
| Time_2 | 2.46 | 11.3 | 0.841 | |
| TF | 2.62 | 11.4 | 0.857 | |
| TF_1 | 2.58 | 11.3 | 0.853 | |
| TF_2 | 2.57 | 11.3 | 0.852 | |
| ER speech | LAEC | 1.61 | −2.05 | 0.697 |
| Time | 2.68 | 11.7 | 0.892 | |
| Time_1 | blue2.70 | blue12.0 | blue0.894 | |
| Time_2 | blue2.75 | blue12.5 | blue0.899 | |
| TF | 2.77 | 11.3 | 0.904 | |
| TF_1 | blue2.80 | blue11.9 | blue0.905 | |
| TF_2 | blue2.88 | blue12.4 | blue0.912 | |
| ER music | LAEC | 1.70 | −1.12 | 0.730 |
| Time | 2.75 | 12.6 | 0.900 | |
| Time_1 | blue2.76 | blue12.8 | blue0.901 | |
| Time_2 | blue2.80 | blue13.0 | blue0.906 | |
| TF | 2.79 | 11.9 | 0.907 | |
| TF_1 | blue2.83 | blue12.3 | blue0.908 | |
| TF_2 | blue2.91 | blue12.6 | blue0.914 | |
| LL speech | LAEC | 1.95 | 1.67 | 0.806 |
| Time | 3.00 | 15.6 | 0.932 | |
| Time_1 | 3.00 | 15.8 | 0.933 | |
| Time_2 | 3.03 | 16.1 | 0.935 | |
| TF | 3.02 | 15.3 | 0.938 | |
| TF_1 | 3.08 | 15.8 | 0.939 | |
| TF_2 | 3.13 | 16.1 | 0.943 | |
| LL music | LAEC | 1.97 | 2.16 | 0.820 |
| Time | 3.07 | 16.0 | 0.935 | |
| Time_1 | 3.03 | 16.1 | 0.936 | |
| Time_2 | 3.04 | 16.2 | 0.937 | |
| TF | 3.12 | 15.8 | 0.944 | |
| TF_1 | 3.17 | 16.1 | 0.944 | |
| TF_2 | 3.18 | 16.2 | 0.946 |