Views
No views yet
1from transformers import pipeline
2import librosa
3
4 #load model
5pipe = pipeline("antispoofing", model="Speech-Arena-2025/DF_Arena_1B_V_1", trust_remote_code=True, device='cuda')
6audio, sr = librosa.load("sample.wav", sr=16000)
7result = pipe(audio)
8print(result)
9# Output:
10{'label': 'spoof', 'logits': [[1.5515458583831787, -1.2254822254180908]], 'score': 0.9414217472076416, 'all_scores': {'spoof': 0.9414217472076416, 'bonafide': 0.05857823044061661}}| Dataset | EER (%) | F1-score | Accuracy (%) |
|---|---|---|---|
| dfadd | 0.00 | 0.9993 | 99.97 |
| add_2023_round_2 | 11.54 | 0.9188 | 88.46 |
| codecfake | 8.37 | 0.8695 | 91.63 |
| asvspoof_2021_la | 4.66 | 0.8037 | 95.34 |
| in_the_wild | 0.91 | 0.9928 | 99.10 |
| asvspoof_2019 | 1.14 | 0.9473 | 98.86 |
| add_2022_track_1 | 22.21 | 0.6678 | 77.79 |
| fake_or_real | 2.92 | 0.9711 | 97.11 |
| asvspoof_2024 | 17.25 | 0.6615 | 82.75 |
| add_2022_track_3 | 2.20 | 0.9357 | 97.80 |
| add_2023_round_1 | 5.08 | 0.9639 | 94.92 |
| librisevoc | 0.15 | 0.9958 | 99.84 |
| asvspoof_2021_df | 1.75 | 0.7577 | 98.25 |
| sonar | 1.09 | 0.9903 | 98.89 |
| Average | 5.919 | 0.8863 | 94.079 |
| Pooled | 9.52 | 0.81 | 90.47 |
1@misc{kulkarni2026compactsslbackbonesmatter,
2 title={Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR},
3 author={Ajinkya Kulkarni and Sandipana Dowerah and Atharva Kulkarni and Tanel Alumäe and Mathew Magimai Doss},
4 year={2026},
5 eprint={2603.06164},
6 archivePrefix={arXiv},
7 primaryClass={cs.SD},
8 url={https://arxiv.org/abs/2603.06164},
9}