tropa-voice-mini: Speech Anti-Spoofing Arena wrapper

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

No paper: the model sits in the Arena's unpublished tier (shown, not ranked).

tropa-voice-mini (wasitaigenerated.com, 2026): the open-weights AI-voice detector behind the wasitaigenerated API's small tier. A wav2vec 2.0 XLS-R 300M encoder fine-tuned end-to-end, with a multi-layer attention head (SLS head): softmax gate over all 25 hidden states, LayerNorm + Linear(1024,256) + GELU, attentive statistics pooling, 2-layer MLP to one AI logit. Per-window zero-mean / unit-std waveform normalisation inside the model. Authors' released checkpoint; training data undisclosed. Input: 16 kHz, first 64600 samples (Arena 4 s cap), shorter clips repeat-padded (the authors score 6 s windows with a 3 s hop over the whole clip and calibrate the top-30 % mean logit). score = -(pre-calibration AI logit), higher = human. 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% 0.72 on ASVspoof2019_LA EER% 3.22 on ASVspoof2021_LA EER% 2.14 on ASVspoof2021_DF EER% 12.19 on ASVspoof5 EER% 1.27 on InTheWild EER% 3.93 on ADD22_eval_31 EER% 5.33 on ADD2023_track12_test_r1 EER% 10.71 on CFAD EER% 0.01 on CD-ADD EER% 5.53 on CVoiceFake_small EER% 5.27 on DeepVoice EER% 0.54 on LibriSeVoc EER% 0.35 on SONAR EER% 3.56 on ArAD EER% 3.73 on DECRO EER% 0.12 on J-SPAW_LA EER% 6.26 on ODSS EER% 0.38 on HABLA EER% 0.1 on DFADD EER% 0.27 on PyAra EER% 3.34 on XMAD EER% 1.17 on EmoFake_test SRR COMPLEMENT 2.83 on EmoSpoofTTS SRR COMPLEMENT 2.93 on LRLspoof

Downloads last month
212
Safetensors
Model size
0.3B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support