List your company

Speaker Verification on VoxCeleb1-O Cleaned (Original)

0.37EER (%)ReDimNet-B6_SF2
0.03282.30894.5856.8611Jun 18, 2019Aug 8, 2020Sep 30, 2021Nov 22, 2022Jan 13, 2024Mar 6, 2025Apr 28, 2026

Evaluation results

MethodLinks
ReDimNet-B6_SF2Params=15.0M, GMACs=20.27, LM=true, AS-Norm=true2024.070.370.03——
LAP (h=16) + ASTPFrontend=LARGE, Param (M)=+ 2.32025.120.37—0.0590.029
ECAPA-TDNN (C=512)Frontend=LARGE, Param (M)=+ 8.6*2025.120.38———
ReDimNet-B5_SF2Params=9.2M, GMACs=9.87, LM=true, AS-Norm=true2024.070.390.037——
ReDimNet-B6_SF2Params=15.0M, GMACs=20.27, LM=true, AS-Norm=false2024.070.40.033——
ReDimNet-B5_SF2Params=9.2M, GMACs=9.87, LM=true, AS-Norm=false2024.070.430.039——
ReDimNet-B4Params=6.3M, GMACs=4.80, LM=true, AS-Norm=true2024.070.440.042——
ECAPA2Params=27.1M, GMACs=187.00*, LM=true, AS-Norm=false2024.070.440.041——
ReDimNet-B3Params=3.0M, GMACs=3.00, LM=true, AS-Norm=true2024.070.470.042——
MHFA (64 heads)Frontend=LARGE, Param (M)=+ 2.222025.120.49—0.0910.045
ReDimNet-B3Params=3.0M, GMACs=3.00, LM=true, AS-Norm=false2024.070.50.063——
ReDimNet-B4Params=6.3M, GMACs=4.80, LM=true, AS-Norm=false2024.070.510.052——
ReDimNet-B2_SF2Params=4.7M, GMACs=0.90, LM=true, AS-Norm=true2024.070.520.06——
ResNet293Params=23.8M, GMACs=28.10, LM=false, AS-Norm=true2024.070.530.057——
ReDimNet-B2_SF2Params=4.7M, GMACs=0.90, LM=true, AS-Norm=false2024.070.570.054——
DF-ResNet233Params=12.3M, GMACs=11.17, LM=false, AS-Norm=true2024.070.580.044——
MHFA (64 heads)Frontend=BASE+, Param (M)=+ 2.22025.120.59—0.0690.041
Gemini DF-ResNet183Params=9.2M, GMACs=8.25, LM=false, AS-Norm=true2024.070.60.064——
LAP (h=12) + ASTPFrontend=BASE+, Param (M)=+ 1.72025.120.61—0.0910.05
DBE (C=512)Frontend=BASE+, Param (M)=+ 14.3*2025.120.63—0.077—
Gemini DF-ResNet114Params=6.5M, GMACs=5.00, LM=false, AS-Norm=true2024.070.690.067——
CAM++Params=7.2M, GMACs=1.15, LM=true, AS-Norm=false2024.070.710.109——
ReDimNet-B1Params=2.2M, GMACs=0.54, LM=true, AS-Norm=true2024.070.730.071——
LAP (h=12) + ASTPFrontend=WavLM BASE, Param (M)=+ 1.72025.120.78—0.0870.056
NeXt-TDNN (C=256, B=3)Params=7.1M, GMACs=1.35*, LM=false, AS-Norm=true2024.070.790.087——
NeXt-TDNN (C=256, B=3)Frontend=Fbank-80, Param (M)=7.12025.120.79—0.087—
NeXt-TDNN-1 (C=256, B=3)Params=6.0M, GMACs=1.13*, LM=false, AS-Norm=true2024.070.810.091——
ResNet34Params=6.6M, GMACs=4.55, LM=true, AS-Norm=false2024.070.820.08——
ECAPA-TDNN (C=512)Frontend=BASE+, Param (M)=+ 8.0*2025.120.84———
ReDimNet-B1Params=2.2M, GMACs=0.54, LM=true, AS-Norm=false2024.070.850.076——
ECAPA-TDNN (C=512)†Frontend=WavLM BASE, Param (M)=+ 8.02025.120.85—0.1190.062
ECAPA-TDNN (C=1024)Frontend=Fbank-80, Param (M)=14.72025.120.87—0.107—
MHFAFrontend=WavLM BASE, Param (M)=+ 0.42025.120.89—0.0940.056
ECAPA (C=512)Params=6.4M, GMACs=1.05, LM=false, AS-Norm=true2024.070.940.092——
Gemini DF-ResNet60Params=4.1M, GMACs=2.50*, LM=false, AS-Norm=true2024.070.940.089——
DF-ResNet56Params=4.5M, GMACs=2.66, LM=false, AS-Norm=true2024.070.960.103——
ECAPA (C=1024)Params=14.9M, GMACs=2.67, LM=true, AS-Norm=false2024.070.980.105——
NeXt-TDNN (C=128, B=3)Params=1.9M, GMACs=0.35*, LM=false, AS-Norm=true2024.071.030.095——
ReDimNet-B0Params=1.0M, GMACs=0.43, LM=true, AS-Norm=true2024.071.070.098——
NeXt-TDNN-1 (C=128, B=3)Params=1.6M, GMACs=0.29*, LM=false, AS-Norm=true2024.071.10.108——
ReDimNet-B0Params=1.0M, GMACs=0.43, LM=false, AS-Norm=false2024.071.160.101——
x-vector†Frontend=LARGE, Param (M)=315.5 + 7.02025.121.52—0.3040.167
x-vector†Frontend=BASE+, Param (M)=94.4 + 6.42025.121.82—0.4110.193
AAM-softmax (x-vector)Model=TDNN (x-vector), Loss=AAM-softmax, Training set=VoxCeleb1 + VoxCeleb22019.062.238———
x-vector†Frontend=WavLM BASE, Param (M)=94.4 + 6.42025.122.25—0.3730.225
Diff-SV2026.042.35———
UF-EMA2026.042.55———
VoiceID2026.042.61———
NDAL2026.042.63———
AAM-softmax (x-vector)Model=TDNN (x-vector), Loss=AAM-softmax, Training set=VoxCeleb22019.062.694———
ExU-Net2026.042.76———
NDML2026.042.9———
Baseline2026.043———
Cai et al.2026.043.12———
Xie et al.Model=Thin ResNet-34, Loss=Softmax, Training set=VoxCeleb22019.063.22———
Okabe et al.Model=TDNN (x-vector), Loss=Softmax, Training set=VoxCeleb12019.063.85———
x-vector†Frontend=Fbank-80, Param (M)=4.62025.123.86—0.3830.23
Chung et al.Model=ResNet-50, Loss=Softmax + Contrastive, Training set=VoxCeleb22019.064.19———
Hajibabaei et al.Model=ResNet-20, Loss=AM-Softmax, Training set=VoxCeleb12019.064.3———
Cai et al.Model=ResNet-34, Loss=A-Softmax, Training set=VoxCeleb12019.064.4———
Nagrani et al.Model=GMM-UBM (i-vector), Training set=VoxCeleb12019.068.8———