WST-Graph-Q82: Speech Anti-Spoofing Arena wrapper

Packaged for the Speech Anti-Spoofing Arena from the authors' code and paper.

WST-Graph (Ng et al., 2026), configuration Q82: a fixed, parameter-free Kymatio wavelet scattering transform (J=8, Q=(8, 2), first + second order, oversampling 1) with log-modulation path normalisation, arranged as a modulation-carrier grid (OrderGrid), length-aware attentive pooling to 64 relative-time bins (ALAP), a channel adapter + depthwise-separable latent blocks and an AASIST graph back-end (GraphAASIST). Trained on ASVspoof 2019 LA train (focal loss, 12 epochs), training seed 9 (the paper names no headline seed; seed 9 is the model card's example). Authors' traced TorchScript program, run as released. Input as the model card: 16 kHz, no normalisation, first 64000 samples (4 s), right zero-pad, input_length = retained sample count. score = logit[bona fide] - logit[spoof] (logits[:,0]-logits[:,1]). FP32 (TF32 off).

Arena results

arena tier arena rank

EER% (lower is better); EmoSpoofTTS and LRLspoof are spoof-only and use 1-SRR% at the model's DeepVoice EER threshold. Per-dataset scores are under .eval_results/.

EER% 2.93 on ASVspoof2019_LA EER% 14.15 on ASVspoof2021_LA EER% 18.97 on ASVspoof2021_DF EER% 35.55 on ASVspoof5 EER% 44.52 on InTheWild EER% 30.43 on ADD22_eval_31 EER% 47.86 on ADD2023_track12_test_r1 EER% 44.95 on CFAD EER% 47.7 on CD-ADD EER% 49.2 on CVoiceFake_small EER% 48.25 on DeepVoice EER% 33.4 on LibriSeVoc EER% 39.37 on SONAR EER% 44.98 on ArAD EER% 27.13 on DECRO EER% 43.64 on J-SPAW_LA EER% 47.62 on ODSS EER% 37.96 on HABLA EER% 21.53 on DFADD EER% 23.34 on PyAra EER% 38.04 on XMAD EER% 36.03 on EmoFake_test SRR COMPLEMENT 30.81 on EmoSpoofTTS SRR COMPLEMENT 36.63 on LRLspoof

Citation

@article{ng2026wstgraph,
  title   = {WST-Graph: Topology-Preserving Wavelet Scattering Front-End for Speech Deepfake Detection},
  author  = {Ng, Kwok-Ho and Song, Tingting and Feng, Bingwen and Xia, Zhihua},
  journal = {arXiv preprint arXiv:2609.29372},
  year    = {2026}
}
Downloads last month
208
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Paper for SpeechAntiSpoofingBenchmarks/WST-Graph-Q82