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 |
| Â | Time | 2.61 | 12.3 | 0.866 |
| Â | Time_1 | 2.56 | 12.2 | 0.865 |
| Â | Time_2 | 2.57 | 12.1 | 0.864 |
| Â | TF | 2.75 | 12.4 | 0.880 |
| Â | TF_1 | 2.70 | 12.4 | 0.875 |
| Â | TF_2 | 2.69 | 12.3 | 0.875 |
Artificial music | LAEC | 1.48 | −2.90 | 0.634 |
| Â | Time | 2.50 | 11.5 | 0.842 |
| Â | Time_1 | 2.44 | 11.4 | 0.841 |
| Â | Time_2 | 2.46 | 11.3 | 0.841 |
| Â | TF | 2.62 | 11.4 | 0.857 |
| Â | TF_1 | 2.58 | 11.3 | 0.853 |
| Â | TF_2 | 2.57 | 11.3 | 0.852 |
ER speech | LAEC | 1.61 | −2.05 | 0.697 |
| Â | Time | 2.68 | 11.7 | 0.892 |
| Â | Time_1 | blue2.70 | blue12.0 | blue0.894 |
| Â | Time_2 | blue2.75 | blue12.5 | blue0.899 |
| Â | TF | 2.77 | 11.3 | 0.904 |
| Â | TF_1 | blue2.80 | blue11.9 | blue0.905 |
| Â | TF_2 | blue2.88 | blue12.4 | blue0.912 |
ER music | LAEC | 1.70 | −1.12 | 0.730 |
| Â | Time | 2.75 | 12.6 | 0.900 |
| Â | Time_1 | blue2.76 | blue12.8 | blue0.901 |
| Â | Time_2 | blue2.80 | blue13.0 | blue0.906 |
| Â | TF | 2.79 | 11.9 | 0.907 |
| Â | TF_1 | blue2.83 | blue12.3 | blue0.908 |
| Â | TF_2 | blue2.91 | blue12.6 | blue0.914 |
LL speech | LAEC | 1.95 | 1.67 | 0.806 |
| Â | Time | 3.00 | 15.6 | 0.932 |
| Â | Time_1 | 3.00 | 15.8 | 0.933 |
| Â | Time_2 | 3.03 | 16.1 | 0.935 |
| Â | TF | 3.02 | 15.3 | 0.938 |
| Â | TF_1 | 3.08 | 15.8 | 0.939 |
| Â | TF_2 | 3.13 | 16.1 | 0.943 |
LL music | LAEC | 1.97 | 2.16 | 0.820 |
| Â | Time | 3.07 | 16.0 | 0.935 |
| Â | Time_1 | 3.03 | 16.1 | 0.936 |
| Â | Time_2 | 3.04 | 16.2 | 0.937 |
| Â | TF | 3.12 | 15.8 | 0.944 |
| Â | TF_1 | 3.17 | 16.1 | 0.944 |
| Â | TF_2 | 3.18 | 16.2 | 0.946 |