AMSDF

Audio Multi-view Spoofing Detection Framework (AMSDF) (Wu et al., IEEE TIFS 2024). AMSDF captures intra- and inter-view spoofing cues from audio–text–emotion correlations: an XLS-R (wav2vec2-xls-r-300m) text view, an ACRNN emotion view (logfbank + delta + delta²), and a raw-audio graph view, fused by heterogeneous graph attention with a group-based readout. The model takes a raw speech waveform (and a derived emotion spectrogram) and returns a score where higher = more bona fide.

• Original code: https://github.com/ItzJuny/AMSDF
• Paper: https://ieeexplore.ieee.org/document/10605761 (IEEE TIFS 2024)
• Original authors: Junyan Wu, Qilin Yin, Ziqi Sheng, Wei Lu, Jiwu Huang, Bin Li
• Parameters: ~325.4 M
• Checkpoint: M1.pth (the authors' best-EER seed; M2/M3 also released)

This repo is self-contained for inference: the multi-view network is vendored in models/, the builder in _net.py, and the exact wrapper in amsdf.py.

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/.

Architecture & inference notes

• Two inputs computed by the wrapper from each waveform: the raw waveform (text + audio views) and an emotion spectrogram (3, 300, 40) = logfbank(40) + Δ + Δ² (first 300 frames), as in the upstream data_io.extract_mel.
• Score = P(real) − P(fake) (out[:,0] − out[:,1]). The repo's label convention is class 0 = real, class 1 = fake, so the bonafide score is the negative of the repo's reported out[:,1].
• Input: raw float32 mono @ 16 kHz, first-64600-sample (≈4 s) window (tile-repeat if shorter). FP32.
• Requires the public XLS-R xlsr2_300m.pt and the emotion model ser_acrnn.pth in pretrained_models/ (loaded at construction; the M*.pth checkpoint then overwrites the fine-tuned weights). The fairseq XLS-R is loaded via merge_with_parent (cross-fairseq-version fix).

Credits

All model design and weights are the work of the original AMSDF authors (https://github.com/ItzJuny/AMSDF). This repository only packages their released checkpoint for the Speech Anti-Spoofing Arena.

@article{wu2024audio,
  author  = {Wu, Junyan and Yin, Qilin and Sheng, Ziqi and Lu, Wei and Huang, Jiwu and Li, Bin},
  journal = {IEEE Transactions on Information Forensics and Security},
  title   = {Audio Multi-view Spoofing Detection Framework Based on Audio-Text-Emotion Correlations},
  year    = {2024}, volume = {19}, pages = {7133--7146},
  doi     = {10.1109/TIFS.2024.3431888}
}