DPDA-GA-Mamba — Speech Anti-Spoofing Arena wrapper

Truong et al., ICASSP 2026 — DPDA-GA: dual-path data-augmented training with gradient alignment (PCGrad). Three back-ends released; this card is one of them (DPDA-GA-AASIST, DPDA-GA-TCM, DPDA-GA-Mamba).

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

DPDA-GA-Mamba: the DPDA-GA training recipe (dual-path data-augmented training with PCGrad gradient alignment between the clean and RawBoost-augmented paths) applied to the XLSR-Mamba architecture -- a fully fine-tuned wav2vec 2.0 XLS-R 300M front-end (last layer) projected to 144-d (Linear + BatchNorm + SELU), followed by a dual-column bidirectional Mamba (v1 selective SSM) back-end: 6 forward + 6 backward Mamba blocks with fused add+RMSNorm, per-direction attention pooling, concatenation, linear fusion and a 2-way head. Official pcgrad/xlsr_mamba checkpoints (trained on ASVspoof2019 LA train), 5-checkpoint weight average as computed by the authors' main.py. Not to be confused with XLSR-MamBo (arXiv 2601.02944).

• Checkpoint (commit-pinned): dpda_ga_mamba_avg5.pth
• Params: 319.3275 M. Trained on ASVspoof2019 LA train only.
• Score: logits[:, 1] (higher = bona fide). Input: first 66,800 samples @16 kHz, shorter clips repeat-padded; FP32. J-SPAW_LA was scored with the PyTorch wrapper; the other datasets with the TensorRT drop-in class (FP32), 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

@inproceedings{dpda_ga,
  author    = {Truong, Duc-Tuan and Liu, Tianchi and Li, Junjie and Tao, Ruijie and Lee, Kong Aik and Chng, Eng Siong},
  title     = {Addressing Gradient Misalignment in Data-Augmented Training for Robust Speech Deepfake Detection},
  booktitle = {ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year      = {2026},
  pages     = {16252--16256},
  doi       = {10.1109/ICASSP55912.2026.11464739}
}