Dialogue Localization on Dialogue localization 100 samples (test)
2.77U-MSE (mm^-1)KM-Speaker
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| KM-Speakermodel_configuration=full proposed model2026.06 | 2.77 | 1.02 | |
| KM-Speakerdisentanglement_masks=MeshTalk disentanglement masks2026.06 | 2.81 | 1.22 | |
| KM-Speakerdisentanglement_strategy=none2026.06 | 2.83 | 1.32 | |
| KM-Speakerencoder=MeshTalk full-face encoder2026.06 | 5.64 | 1.42 | |
| MeshTalk2026.06 | 6.6 | 1.68 |