Add Pellav2 submission for LRLspoof (1-SRR=4.0900)

#35
by korallll - opened
Files changed (1) hide show
  1. submissions/pellav2.yaml +53 -0
submissions/pellav2.yaml ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ schema_version: 4
2
+ system:
3
+ name: Pellav2
4
+ slug: pellav2
5
+ description: 'Pellav2 (Sadanie, August 2026): an open-weights audio deepfake detector. A w2v-BERT 2.0
6
+ conformer encoder (24 layers, fine-tuned) on 80-bin kaldi log-mel fbank features (x32768, per-utterance
7
+ mean/variance normalised, frame pairs stacked to 160-d, trailing digital silence trimmed), a softmax-weighted
8
+ mix of all 25 hidden states, masked mean pooling and Linear(1024,1) -> logit(fake). Authors'' released
9
+ checkpoint; training data undisclosed. Input as the authors'' pellav2_infer.py: 16 kHz, mean removal,
10
+ 20 Hz high-pass (zero-phase 2nd-order Butterworth), centre 4 s (64000-sample) crop, ITU-R BS.1770
11
+ loudness normalisation to -23 LUFS (pyloudnorm semantics), shorter clips zero-padded. score = -logit(fake),
12
+ higher = bona fide. FP32 (TF32 off).'
13
+ code: https://huggingface.co/Sadanie/pellav2-audio-deepfake-detector/blob/b668eb4dfe4719665692c73c1d70383680afff6b/pellav2_infer.py
14
+ checkpoint: https://huggingface.co/SpeechAntiSpoofingBenchmarks/Pellav2/blob/0de5ef2278e4fb485bd57199a13d15cbf3686954/pellav2_detector.safetensors
15
+ params_millions: 580.493146
16
+ dataset:
17
+ id: lab260/LRLspoof
18
+ revision: ac76f0e5472fdb68a6726e1fa94a148316e76916
19
+ split: test
20
+ scores:
21
+ srr_complement: 4.090037410795688
22
+ n_trials: 1304455
23
+ n_skipped: 286
24
+ calibration:
25
+ source_dataset: SpeechAntiSpoofingBenchmarks/DeepVoice
26
+ threshold: 3.9936474915254236
27
+ artifact:
28
+ scores_url: https://huggingface.co/SpeechAntiSpoofingBenchmarks/Pellav2/resolve/a4c57c5c7d59e25373da872813e5af321373c8f7/.eval_results/lab260/LRLspoof/scores.txt
29
+ scores_sha256: cda717db2cd95dd1e2421f0e1c1cf8d291a86ddb1a8f6d80a3a7d7219f3f5b12
30
+ bench_version: speech-spoof-bench==0.4.1
31
+ reproduction:
32
+ reproduced_by: SpeechAntiSpoofingBenchmarks
33
+ reproduced_at: '2026-11-14'
34
+ reproduced_bench_version: speech-spoof-bench==0.4.1
35
+ match: inference
36
+ submitter:
37
+ hf_username: korallll
38
+ contact: k.n.borodin@mtuci.ru
39
+ submitted_at: '2026-11-14'
40
+ notes: 'Pellav2 = HF Sadanie/pellav2-audio-deepfake-detector @ b668eb4dfe4719665692c73c1d70383680afff6b,
41
+ pellav2_detector.pt (sha256 d21aedfb52e2f87706c2753715a29665482b0f4a034800c31579df5d458559ac; 776 fp32
42
+ tensors, loaded with weights_only=True, strict, all used) re-hosted unchanged as pellav2_detector.safetensors
43
+ (sha256 050234557deabc95d9697a2ede42a5152ed06ba500a4488a42542ab793af428f). MIT per the model card (no
44
+ LICENSE file). Model class and preprocessing vendored from the repo''s pellav2_infer.py (sha256 bf13cc97...,
45
+ read, never imported from the Hub cache); the w2v-BERT 2.0 backbone is built from the facebook/w2v-bert-2.0
46
+ config (@ da985ba0) and every weight comes from the checkpoint. pyloudnorm''s integrated loudness is
47
+ re-implemented (same filters, gating and op order; not installed here) with the script''s RMS fallback
48
+ for clips < 0.4 s. Verified against the authors'' Detector + load_clip on 19 J-SPAW_LA-derived clips:
49
+ identical preprocessing, max |d score| 2.8e-6. The authors'' own window is a centre 4 s crop with zero-padding,
50
+ so it is kept (no Arena first-64600/repeat-pad deviation). Training data is undisclosed (no paper, card
51
+ silent) -> no trained_on entries, but leakage cannot be ruled out: J-SPAW_LA EER is 0.0% with saturated,
52
+ perfectly separated scores. The batch shortlist cites a DFADD EER of 1.9% (not on the pinned card).
53
+ No paper -> unpublished tier.'