Audio Spoofing Detection on ASVspoof Logical Access 2019 (Evaluation)
0.83EERAASIST
Evaluation results
| Method | Links | ||
|---|---|---|---|
| AASIST# Param=297K, Front-end=Raw waveform, System=Ours2021.10 | 0.83 | 0.0275 | |
| AASIST#params=297K, GPU=Yes, Front-end=Raw audio2024.03 | 0.83 | — | |
| Wave2Vec 2.0 - SVM#params=808, GPU=No, Front-end=Raw audio2024.03 | 0.9 | — | |
| AASIST-L# Param=85K, Front-end=Raw waveform, System=Ours2021.10 | 0.99 | 0.0309 | |
| AASIST-L#params=85K, GPU=Yes, Front-end=Raw audio2024.03 | 0.99 | — | |
| RawGAT-ST# Param=437K, Front-end=Raw waveform, System=Tak et al. [11]2021.10 | 1.06 | 0.0335 | |
| RawGAT-ST#params=437K, GPU=Yes, Front-end=Raw audio2024.03 | 1.06 | — | |
| wav2vec 2.0-light-DARTS#params=NA, GPU=Yes, Front-end=Raw audio2024.03 | 1.09 | — | |
| Wave2Vec 2.0 - MLP#params=77K, GPU=No, Front-end=Raw audio2024.03 | 1.11 | — | |
| SENet# Param=1,100K, Front-end=FFT, System=Zhang et al. [28]2021.10 | 1.14 | 0.0368 | |
| SENet#params=1,100K, GPU=Yes, Front-end=FFT2024.03 | 1.14 | — | |
| Wave2Vec 2.0 - Log#params=769, GPU=No, Front-end=Raw audio2024.03 | 1.25 | — | |
| Res-TSSDNet# Param=350K, Front-end=Raw waveform, System=Hua et al. [29]2021.10 | 1.64 | 0.0481 | |
| Res-TSSDNet#params=350K, GPU=Yes, Front-end=Raw audio2024.03 | 1.64 | — | |
| Raw PC-DARTS# Param=24,480K, Front-end=Raw waveform, System=Ge et al. [30]2021.10 | 1.77 | 0.0517 | |
| Raw PC-DARTS#params=24,480K, GPU=Yes, Front-end=Raw audio2024.03 | 1.77 | — | |
| MCG-Res2Net50# Param=960K, Front-end=CQT, System=Li et al. [31]2021.10 | 1.78 | 0.052 | |
| ResNet18-LMCL-FMFront-end=LFB, System=Chen et al. [32]2021.10 | 1.81 | 0.052 | |
| LCNN-LSTM-sum# Param=276K, Front-end=LFCC, System=Wang et al. [33]2021.10 | 1.92 | 0.0524 | |
| LCNN-LSTM-sum#params=276K, GPU=Yes, Front-end=LFCC2024.03 | 1.92 | — | |
| Capsule networkFront-end=LFCC, System=Luo et al. [34]2021.10 | 1.97 | 0.0538 | |
| Resnet18-OC-softmaxFront-end=LFCC, System=Zhang et al. [35]2021.10 | 2.19 | 0.059 | |
| SE-Res2Net50Front-end=CQT, System=Li et al. [36]2021.10 | 2.5 | 0.0743 | |
| LCNN-Dual attentionFront-end=LFCC, System=Ma et al. [37]2021.10 | 2.76 | 0.0777 | |
| GMMFront-end=LFCC, System=Tak et al. [9]2021.10 | 3.5 | 0.0904 | |
| ResNet18-GAT-TFront-end=LFB, System=Tak et al. [10]2021.10 | 4.71 | 0.0894 | |
| PC-DARTS# Param=7,510K, Front-end=LFCC, System=Ge et al. [38]2021.10 | 4.96 | 0.0914 | |
| ResNet-34 adaptationFeature=Mel-spectrogram, Architecture=Residual Neural Network2020.08 | 5.32 | 0.1514 | |
| LFCC - GMMBaseline=ASVspoof 2019 Organizer Baseline2020.08 | 8.09 | 0.2116 | |
| CQCC - GMMBaseline=ASVspoof 2019 Organizer Baseline2020.08 | 9.57 | 0.2366 |