Nes2Net-X-CtrSVDD: Speech Anti-Spoofing Arena wrapper

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

Singing-voice deepfake detector: Nes2Net-X trained on CtrSVDD (controlled singing voice deepfake detection, SVDD 2024), not on speech. A fully fine-tuned WavLM-Large front-end whose 25 hidden states are merged by squeeze-excitation layer attention (SEA), followed by the Nes2Net-X nested Res2Net TDNN back-end (dilation 1), mean temporal pooling and a single logit (label 1 = real). Authors' WavLM_Nes2Net_X seed-420 epoch-75 checkpoint, FP32, deterministic first-64600-sample window (repeat-padded if shorter). Included as an out-of-domain (singing-voice) reference.

• Weights: not re-hosted (the authors grant no license for them). Get the authors' checkpoint; its sha256 is in the meta.yaml notes.
• Params: 315.964534 M. Trained on CtrSVDD (singing-voice deepfakes) train; a singing-voice model scored on speech.
• Score: higher = bona fide. J-SPAW_LA was scored with the PyTorch wrapper; the other datasets with the TensorRT drop-in class (FP32 on the RTX 4070 Ti SUPER; FP32 on the RTX 5060 Ti), which matched PyTorch on the full J-SPAW_LA set.

Arena results

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/.

Citation

@article{Nes2Net,
  author={Liu, Tianchi and Truong, Duc-Tuan and Das, Rohan Kumar and Lee, Kong Aik and Li, Haizhou},
  journal={IEEE Transactions on Information Forensics and Security},
  title={Nes2Net: A Lightweight Nested Architecture for Foundation Model Driven Speech Anti-Spoofing},
  year={2025},
  volume={20},
  pages={12005--12018},
  doi={10.1109/TIFS.2025.3626963}
}