List your company
ResearchFormerly WizWand

35-way Speech Classification on Speech Commands 16kHz 35-way (test)

96.82AccuracyMRConv, Fourier
59.816869.423479.0388.6366Aug 9, 2022Dec 10, 2022Apr 12, 2023Aug 14, 2023Dec 15, 2023Apr 16, 2024Aug 18, 2024
Updated 8mo ago

Evaluation results

MethodLinks
MRConv, FourierSampling rate=16 kHz, Input length=16,0002024.0896.82
MRConv, FourierArchitecture Class=Convolutional Models, Parameterization=Fourier2024.0896.82
Liquid-S4Parameters=224K, Input length=160002022.0896.78
Liquid-S4Sampling rate=16 kHz, Input length=16,0002024.0896.78
Liquid-S4Architecture Class=State Space Models2024.0896.78
S5Parameters=280K, Input length=160002022.0896.52
S5Sampling rate=16 kHz, Input length=16,0002024.0896.52
S5Architecture Class=State Space Models2024.0896.52
MRConv, DilatedSampling rate=16 kHz, Input length=16,0002024.0896.47
MRConv, DilatedArchitecture Class=Convolutional Models, Parameterization=Dilated2024.0896.47
SGConvSampling rate=16 kHz, Input length=16,0002024.0896.42
SGConvArchitecture Class=Convolutional Models2024.0896.42
S4-LegSParameters=307K, Input length=160002022.0896.08
S4-LegSSampling rate=16 kHz, Input length=16,0002024.0896.08
S4-LegSArchitecture Class=State Space Models2024.0896.08
S4D-LegSParameters=306K, Input length=160002022.0895.83
S4D-LegSSampling rate=16 kHz, Input length=16,0002024.0895.83
S4D-LegSArchitecture Class=State Space Models2024.0895.83
ConvNetParameters=26.2M, Input length=160002022.0895.51
ConvNetSampling rate=16 kHz, Input length=16,0002024.0895.51
ConvNetArchitecture Class=CNNs2024.0895.51
MRConv, Fourier+SparseSampling rate=16 kHz, Input length=16,0002024.0895.21
MRConv, Fourier+SparseArchitecture Class=Convolutional Models, Parameterization=Fourier+Sparse2024.0895.21
XResNet-50Parameters=904K, Input length=160002022.0883.01
XResNet-50Sampling rate=16 kHz, Input length=16,0002024.0883.01
XResNet-50Architecture Class=CNNs2024.0883.01
ResNet-1Parameters=216K, Input length=160002022.0877.86
ResNet-18Sampling rate=16 kHz, Input length=16,0002024.0877.86
ResNet-1Architecture Class=CNNs2024.0877.86
InceptionNetParameters=481K, Input length=160002022.0861.24
InceptionNetSampling rate=16 kHz, Input length=16,0002024.0861.24
InceptionNetArchitecture Class=CNNs2024.0861.24