List your company
ResearchFormerly WizWand

Audio-visual speech separation on LRS2-2Mix (test)

16SI-SNRiIIANet
9.7611.381314.62Dec 21, 2022Feb 6, 2023Mar 25, 2023May 11, 2023Jun 27, 2023Aug 13, 2023Sep 29, 2023
Updated 8mo ago

Evaluation results

MethodLinks
IIANet2023.081616.2—
IIANet2023.081616.23.23
IIANetMACs (G)=18.6, Params (M)=3.1, CPU (s)=1.27, GPU (ms)=110.11, GPU Memory (MB)=12.52023.081616.23.23
IIANet-fast2023.0815.115.43.11
IIANet-fastMACs (G)=11.9, Params (M)=3.1, CPU (s)=0.52, GPU (ms)=70.94, GPU Memory (MB)=12.52023.0815.115.43.11
RTFS-Net-12Params (M)=0.7, MACS (G)=56.4, Time (ms)=109.92023.0914.915.13.07
RTFS-Net-6Params (M)=0.7, MACS (G)=30.5, Time (ms)=64.72023.0914.614.83.03
CTCNet2022.1214.4——
CTCNet2023.0814.314.6—
CTCNet2023.0814.314.63.08
CTCNetMACs (G)=167.1, Params (M)=7.0, CPU (s)=1.26, GPU (ms)=84.17, GPU Memory (MB)=75.82023.0814.314.63.08
CTCNetParams (M)=7.0, MACS (G)=167.2, Time (ms)=122.72023.0914.314.63.08
RTFS-Net-4Params (M)=0.7, MACS (G)=21.9, Time (ms)=57.82023.0914.114.33.02
AVLIT-82023.0812.813.1—
AVLIT-82023.0812.813.12.56
AVLITMACs (G)=18.2, Params (M)=5.8, CPU (s)=0.46, GPU (ms)=62.51, GPU Memory (MB)=24.02023.0812.813.12.56
AVLIT-82022.1212.8——
AVLITParams (M)=5.8, MACS (G)=36.4, Time (ms)=53.42023.0912.813.12.56
AVConvTasNet2023.0812.512.8—
AVConvTasNet2023.0812.512.82.69
AVConvTasNetMACs (G)=23.8, Params (M)=16.5, CPU (s)=1.06, GPU (ms)=62.51, GPU Memory (MB)=117.052023.0812.512.82.69
CaffNet-CPhase Information Usage=GT phase + estimated amplitude2022.1212.5——
AV-ConvTasNet2022.1212.5——
CaffNet-C*2023.0912.51.15—
AV-ConvTasnetParams (M)=16.5, Time (ms)=60.32023.0912.512.8—
VisualVoice2022.1211.8——
Thanh-Dat2023.0911.6——
Visualvoice2023.0811.511.82.78
VisualvoiceMACs (G)=9.7, Params (M)=77.8, CPU (s)=2.98, GPU (ms)=110.31, GPU Memory (MB)=313.742023.0811.511.82.78
VisualVoiceParams (M)=77.8, Time (ms)=130.22023.0911.511.82.78
LWTNet2023.0810.8——
CaffNet-C2023.0810—1.15
CaffNet-CPhase Information Usage=Standard2022.1210——