Acoustic Echo Cancellation on AEC Challenge blind 2022 (test)
4.49MOSavgE2E-AEC
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| E2E-AECExperiment=6, Configuration=PL + Knowledge Transfer + Time Alignment + VAD with masking2026.01 | 4.49 | 4.69 | 4.31 | 79.02 | 4.73 | 4.24 | |
| E2E-AECExperiment=4, Configuration=PL + Knowledge Transfer + Time Alignment2026.01 | 4.47 | 4.68 | 4.28 | 50.43 | 4.66 | 4.27 | |
| E2E-AECExperiment=5, Configuration=PL + Knowledge Transfer + Time Alignment + VAD2026.01 | 4.47 | 4.69 | 4.32 | 50.98 | 4.65 | 4.23 | |
| E2E-AECExperiment=3, Configuration=PL + Knowledge Transfer2026.01 | 4.44 | 4.66 | 4.23 | 48.51 | 4.66 | 4.2 | |
| E2E-AECExperiment=2, Configuration=Progressive Learning (PL)2026.01 | 4.4 | 4.6 | 4.16 | 47.35 | 4.68 | 4.16 | |
| E2E-AECExperiment=1, Configuration=Base Model2026.01 | 4.32 | 4.54 | 4.08 | 46.82 | 4.66 | 4.01 | |
| NCA-CRNFramework=End-to-End2026.01 | 4.29 | — | — | — | — | — |