TitaNet-M (NME-SC)

Reported on 4 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Speech4 results

Speaker DiarizationonNIST-SRE 2000
DER(%)· 2021-10-08
6.47
best: 8.39 (x-vector (PLDA + AHC))
TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context arXiv:2110.04410
Speaker DiarizationonCH109
DER(%)· 2021-10-08
1.13
best: 9.72 (x-vector (PLDA + AHC))
TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context arXiv:2110.04410
Speaker DiarizationonAMI MixHeadset
DER(%)· 2021-10-08
1.79
best: 2.22 (TitaNet-S (NME-SC))
TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context arXiv:2110.04410
Speaker DiarizationonAMI Lapel
DER(%)· 2021-10-08
1.99
best: 2.36 (ECAPA (SC))
TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context arXiv:2110.04410