MANNER-S + MV-AT (8.1GF)

Reported on 6 benchmarks across 1 task · 1 paper · 1 SOTA

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

Audio6 results

Speech EnhancementonVoiceBank + DEMAND
Para. (M)· 2022-08-22
1.38
best: 65 (ROSE-CD(PESQ))
SOTA
Multi-View Attention Transfer for Efficient Speech Enhancement arXiv:2208.10367
Speech EnhancementonVoiceBank + DEMAND
CBAK· 2022-08-22
3.61
best: 3.99 (MP-SENet)
Multi-View Attention Transfer for Efficient Speech Enhancement arXiv:2208.10367
Speech EnhancementonVoiceBank + DEMAND
COVL· 2022-08-22
3.82
best: 4.4 (Mamba-SEUNet L (+PCS))
Multi-View Attention Transfer for Efficient Speech Enhancement arXiv:2208.10367
Speech EnhancementonVoiceBank + DEMAND
CSIG· 2022-08-22
4.45
best: 4.82 (Mamba-SEUNet L (+PCS))
Multi-View Attention Transfer for Efficient Speech Enhancement arXiv:2208.10367
Speech EnhancementonVoiceBank + DEMAND
PESQ (wb)· 2022-08-22
3.12
best: 3.99 (ROSE-CD(PESQ))
Multi-View Attention Transfer for Efficient Speech Enhancement arXiv:2208.10367
Speech EnhancementonVoiceBank + DEMAND
STOI· 2022-08-22
95
best: 96.22 (ZipEnhancer (S, \lamba_6 = 0.2))
Multi-View Attention Transfer for Efficient Speech Enhancement arXiv:2208.10367