Audio-visual target speaker extraction on VoxCeleb-2mix impaired visual scenarios (test)
11.27SI-SNRMeMo (TDSE)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MeMo (TDSE)Speaker Bank=VPInit, Evaluation Mode=Offline, Evaluation Setting=Visual+TgtEnro2025.07 | 11.27 | — | — | — | |
| MeMo (BSRNN)#Param (M)=32.23, Macs (G/s)=224.93, Retrained on impaired datasets=false, Config=Contextual bank, Visual+SelfEnro2025.07 | 10.98 | 9.47 | 2.01 | 0.83 | |
| MeMo (TDSE)Speaker Bank=VPInit, Evaluation Mode=Offline, Evaluation Setting=Visual+SelfEnro2025.07 | 10.84 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VPInit, Evaluation Mode=Offline, Evaluation Setting=Visual+PreEnro2025.07 | 10.83 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Offline, Evaluation Setting=Visual+TgtEnro2025.07 | 10.69 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Offline, Evaluation Setting=Visual+PreEnro2025.07 | 10.52 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Offline, Evaluation Setting=Visual+SelfEnro2025.07 | 10.35 | — | — | — | |
| MeMo (TDSE)#Param (M)=23.00, Macs (G/s)=20.72, Retrained on impaired datasets=false, Config=Contextual bank, Visual+SelfEnro2025.07 | 10.34 | 10.76 | 2.08 | 0.84 | |
| TDSEEvaluation Mode=Offline, Evaluation Setting=Visual Only2025.07 | 10.32 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Offline, Evaluation Setting=Visual Only2025.07 | 10.25 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VPInit, Evaluation Mode=Online, Evaluation Setting=Visual+TgtEnro2025.07 | 9.85 | — | — | — | |
| MeMo (USEV)#Param (M)=16.14, Macs (G/s)=18.15, Retrained on impaired datasets=false, Config=Contextual bank, Visual+SelfEnro2025.07 | 9.47 | 9.85 | 1.91 | 0.83 | |
| MoMuSE#Param (M)=26.39, Macs (G/s)=21.51, Retrained on impaired datasets=false2025.07 | 9.46 | 10 | 1.91 | 0.84 | |
| MeMo (TDSE)Speaker Bank=VPInit, Evaluation Mode=Online, Evaluation Setting=Visual+SelfEnro2025.07 | 9.43 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VPInit, Evaluation Mode=Online, Evaluation Setting=Visual+PreEnro2025.07 | 9.24 | — | — | — | |
| ImagineNET#Param (M)=33.61, Macs (G/s)=518.95, Retrained on impaired datasets=false2025.07 | 8.97 | 9.4 | 1.92 | 0.82 | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Online, Evaluation Setting=Visual+TgtEnro2025.07 | 8.76 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Online, Evaluation Setting=Visual+SelfEnro2025.07 | 8.64 | — | — | — | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Online, Evaluation Setting=Visual+PreEnro2025.07 | 8.6 | — | — | — | |
| MuSE#Param (M)=26.13, Macs (G/s)=21.29, Retrained on impaired datasets=false2025.07 | 8.32 | 8.83 | 1.88 | 0.81 | |
| MeMo (TDSE)Speaker Bank=VInit, Evaluation Mode=Online, Evaluation Setting=Visual Only2025.07 | 8.16 | — | — | — | |
| TDSE*#Param (M)=22.15, Macs (G/s)=20.03, Retrained on impaired datasets=true2025.07 | 8.13 | 8.53 | 1.85 | 0.81 | |
| TDSEEvaluation Mode=Online, Evaluation Setting=Visual Only2025.07 | 8.13 | — | — | — | |
| BSRNN*#Param (M)=31.17, Macs (G/s)=224.06, Retrained on impaired datasets=true2025.07 | 7.98 | 8.07 | 1.83 | 0.78 | |
| USEV*#Param (M)=15.29, Macs (G/s)=17.46, Retrained on impaired datasets=true2025.07 | 7.4 | 7.76 | 1.69 | 0.79 | |
| Mixture#Param (M)=02025.07 | 0 | 0.09 | 1.26 | 0.63 | |
| Mixture2025.07 | 0 | — | — | — |