SpatialNet

Reported on 6 benchmarks across 3 tasks

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

Audio5 results

Speech RecognitiononRealMAN
CER
14.5
best: 14.4 (CleanMel-L-mask)
Speech EnhancementonRealMAN
DNSMOS BAK
3.43
best: 4.03 (CleanMel-L-map)
Speech EnhancementonRealMAN
DNSMOS OVRL
2.71
best: 3.25 (CleanMel-L-map)
Speech EnhancementonRealMAN
DNSMOS SIG
3.23
best: 3.55 (CleanMel-L-map)
Speech EnhancementonRealMAN
PESQ-WB
2.16

Speech1 result

Automatic Speech Recognition (ASR)onRealMAN
CER
14.5
best: 14.4 (CleanMel-L-mask)