Speaker Verification on VoxPhrase Eval-1
8.17EER (Top-1%)ERes2Net-L with Neural Re-scoring
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ERes2Net-L with Neural Re-scoring# Params=20.5 M, Vox-O=0.57, Verifier (# Params)=c1024→264 k, TI-Enroll=10 s, TD-Enroll=Phrase2026.06 | 8.17 | 4.89 | 3.74 | 1.37 | 4.54 | |
| ERes2Net-L with Neural Re-scoring# Params=20.5 M, Vox-O=0.57, Verifier (# Params)=c1024→264 k, TI-Enroll=3 s, TD-Enroll=Phrase2026.06 | 8.99 | 5.57 | 4.3 | 1.64 | 5.13 | |
| ERes2Net-L# Params=20.5 M, Vox-O=0.57, Verifier=None, TI-Enroll=10 s, TD-Enroll=None2026.06 | 9.32 | 5.73 | 4.44 | 1.59 | 5.27 | |
| CAM++ with Neural Re-scoring# Params=7.2 M, Vox-O=0.65, Verifier (# Params)=c512→199 k, TI-Enroll=10 s, TD-Enroll=Phrase2026.06 | 9.58 | 5.79 | 4.44 | 1.6 | 5.35 | |
| Ecapa-Tdnn with Neural Re-scoring# Params=20.8 M, Vox-O=0.86, Verifier (# Params)=c3072→526 k, TI-Enroll=10 s, TD-Enroll=Phrase2026.06 | 9.98 | 6.3 | 4.79 | 1.94 | 5.75 | |
| CAM++ with Neural Re-scoring# Params=7.2 M, Vox-O=0.65, Verifier (# Params)=c512→199 k, TI-Enroll=3 s, TD-Enroll=Phrase2026.06 | 10.47 | 6.58 | 5.13 | 1.93 | 6.03 | |
| Ecapa-Tdnn with Neural Re-scoring# Params=20.8 M, Vox-O=0.86, Verifier (# Params)=c3072→526 k, TI-Enroll=3 s, TD-Enroll=Phrase2026.06 | 10.74 | 7.1 | 5.59 | 2.35 | 6.45 | |
| ERes2Net-L# Params=20.5 M, Vox-O=0.57, Verifier=None, TI-Enroll=3 s, TD-Enroll=None2026.06 | 11.02 | 7.17 | 5.64 | 2.21 | 6.51 | |
| CAM++# Params=7.2 M, Vox-O=0.65, Verifier=None, TI-Enroll=10 s, TD-Enroll=None2026.06 | 11.33 | 7.03 | 5.38 | 2.03 | 6.44 | |
| Ecapa-Tdnn# Params=20.8 M, Vox-O=0.86, Verifier=None, TI-Enroll=10 s, TD-Enroll=None2026.06 | 11.37 | 7.21 | 5.59 | 2.2 | 6.59 | |
| ERes2Net-L with Neural Re-scoring# Params=20.5 M, Vox-O=0.57, Verifier (# Params)=c1024→264 k, TI-Enroll=None, TD-Enroll=Phrase2026.06 | 11.94 | 7.91 | 6.36 | 2.68 | 7.22 | |
| ERes2Net-L# Params=20.5 M, Vox-O=0.57, Verifier=None, TI-Enroll=None, TD-Enroll=Phrase (0.8∼3 s)2026.06 | 12.92 | 8.81 | 7.1 | 2.99 | 7.96 | |
| CAM++# Params=7.2 M, Vox-O=0.65, Verifier=None, TI-Enroll=3 s, TD-Enroll=None2026.06 | 13.34 | 9.03 | 7.22 | 3.01 | 8.15 | |
| Ecapa-Tdnn# Params=20.8 M, Vox-O=0.86, Verifier=None, TI-Enroll=3 s, TD-Enroll=None2026.06 | 13.51 | 9.09 | 7.28 | 3.04 | 8.23 | |
| CAM++ with Neural Re-scoring# Params=7.2 M, Vox-O=0.65, Verifier (# Params)=c512→199 k, TI-Enroll=None, TD-Enroll=Phrase2026.06 | 13.59 | 9.17 | 7.39 | 3.09 | 8.31 | |
| Ecapa-Tdnn with Neural Re-scoring# Params=20.8 M, Vox-O=0.86, Verifier (# Params)=c3072→526 k, TI-Enroll=None, TD-Enroll=Phrase2026.06 | 14.63 | 10.02 | 8.33 | 4.08 | 9.27 | |
| CAM++# Params=7.2 M, Vox-O=0.65, Verifier=None, TI-Enroll=None, TD-Enroll=Phrase (0.8∼3 s)2026.06 | 14.73 | 10.1 | 8.14 | 3.62 | 9.15 | |
| Ecapa-Tdnn# Params=20.8 M, Vox-O=0.86, Verifier=None, TI-Enroll=None, TD-Enroll=Phrase (0.8∼3 s)2026.06 | 15.79 | 10.98 | 9.11 | 4.35 | 10.06 |