Speaker Verification on VoxCeleb1 extended (test)
1.07EERH/ASP (AP+softmax)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| H/ASP (AP+softmax)Augmentation=true, Parameters (M)=8.02023.10 | 1.07 | — | — | |
| RecXiAugmentation=true, Parameters (M)=7.06, system_id=#92023.10 | 1.075 | — | 0.114 | |
| MFA-TDNN (standard)Augmentation=true, Parameters (M)=7.322023.10 | 1.083 | — | 0.118 | |
| Xi + tResNet (baseline)Augmentation=true, Parameters (M)=6.54, system_id=#82023.10 | 1.11 | — | 0.125 | |
| MFA-TDNN (lite)Augmentation=true, Parameters (M)=5.932023.10 | 1.138 | — | 0.121 | |
| ECAPA-TDNN (re-implemented)Augmentation=true, Parameters (M)=6.19, system_id=#62023.10 | 1.194 | — | 0.133 | |
| RecXiAugmentation=false, Parameters (M)=7.06, system_id=#52023.10 | 1.197 | — | 0.124 | |
| ECAPA-TDNN (MBFA-MW)Augmentation=true2023.10 | 1.22 | — | 0.135 | |
| Xi + tResNet (baseline)Augmentation=false, Parameters (M)=6.54, system_id=#42023.10 | 1.224 | — | 0.142 | |
| ECAPA-TDNNAugmentation=true, Parameters (M)=6.22023.10 | 1.24 | — | 0.142 | |
| ResNet34 (re-implemented)Augmentation=true, Parameters (M)=6.63, system_id=#72023.10 | 1.252 | — | 0.139 | |
| SI-Net50Augmentation=false2023.10 | 1.28 | — | — | |
| TSP + tResNet (baseline)Augmentation=false, Parameters (M)=6.21, system_id=#32023.10 | 1.281 | — | 0.141 | |
| RSKNet-MTSPAugmentation=false, Parameters (M)=13.92023.10 | 1.3 | — | 0.152 | |
| H/ASP (AAM-softmax)Augmentation=true, Parameters (M)=8.02023.10 | 1.35 | — | — | |
| H/ASP (AP+softmax)Augmentation=false, Parameters (M)=8.02023.10 | 1.42 | — | — | |
| ResNet34 (re-implemented)Augmentation=false, Parameters (M)=6.63, system_id=#22023.10 | 1.423 | — | 0.158 | |
| ECAPA-TDNN (re-implemented)Augmentation=false, Parameters (M)=6.19, system_id=#12023.10 | 1.451 | — | 0.153 | |
| Res2Net-26w8sAugmentation=true, Parameters (M)=9.32023.10 | 1.47 | — | 0.169 | |
| H/ASP (AAM-softmax)Augmentation=false, Parameters (M)=8.02023.10 | 1.56 | — | — | |
| ResNet34-SKDFEAugmentation=false, Parameters (M)=5.982023.10 | 1.59 | — | 0.179 | |
| AAM-softmax (x-vector)Model=TDNN (x-vector), Loss=AAM-softmax, Training set=VoxCeleb22019.06 | 2.762 | — | — | |
| Xie et al.Model=Thin ResNet-34, Loss=Softmax, Training set=VoxCeleb22019.06 | 3.13 | — | — | |
| Chung et al.Model=ResNet-50, Loss=Softmax + Contrastive, Training set=VoxCeleb22019.06 | 4.42 | — | — | |
| ResNet-50Test-time augmentation=(3) Mean of 100 distances2018.06 | 4.42 | 0.524 | — |