ModelPublicadoExterno

MMS 300M AntiDeepfake NDA

Bit-exact copy of nii-yamagishilab/mms-300m-anti-deepfake-nda (the AntiDeepfake collection, nii-yamagishilab), hosted for the Speech Anti-Spoofing Arena. Trained without data augmentation (NDA).

Fuentehuggingface.co/SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA/tree/6d9b482c6f58616253e6be80544912f3c627184a

TipoOtherModelo basenii-yamagishilab/mms-300m-anti-deepfake-ndaVersiónRevision 6d9b482Bcalidad

Fuente del modelo

Tipo: Other
Tamaño del archivo: 1,18 GB
Modelo base: nii-yamagishilab/mms-300m-anti-deepfake-nda
Versión: vRevision 6d9b482
Actualizado: 25 oct 2026
Descargas: 21
Fuente: Hugging Face

Descripción de la fuente

MMS-300M-AntiDeepfake-NDA

Bit-exact copy of nii-yamagishilab/mms-300m-anti-deepfake-nda (the AntiDeepfake collection, nii-yamagishilab), hosted for the Speech Anti-Spoofing Arena. Trained without data augmentation (NDA).

Paper: Post-training for Deepfake Speech Detection, arXiv:2506.21090
Original code: https://github.com/nii-yamagishilab/AntiDeepfake
Backbone: MMS-300M; head = AdaptiveAvgPool1d + Linear(D, 2). Score = real logit.
License: CC BY-NC-SA 4.0 (non-commercial) — inherited from the source.

⚠️ Evaluation restriction

It is STRICTLY PROHIBITED to evaluate this model on its training / in-domain corpora: ASVspoof2019 LA, ASVspoof2021 LA/DF, ASVspoof5, CFAD, DECRO, DFADD, "Diffuse or Confuse", DiffSSD, DSD, HABLA, MLAAD, SpoofCeleb, CVoiceFake, any LibriTTS-based set (incl. SONAR, CD-ADD, LibriSeVoc), any VoxCeleb-based set, WaveFake, any FLEURS-based set, CodecFake (any variant), any AIShell3-based set, and any MLS-based set. On the Arena it is evaluated only on ArAD and In-the-Wild.

All credit to the original authors (Ge, Wang, et al., NII).

Arena results

EER% 0.83 on XMAD EER% 0.64 on PyAra EER% 5.63 on ODSS EER% 0.25 on J-SPAW_LA EER% 5.73 on DeepVoice EER% 3.77 on InTheWild EER% 8.75 on ArAD 1-SRR% 1.26 on LRLspoof EER% 2.17 on EmoFake_test 1-SRR% 12.79 on EmoSpoofTTS arena tier arena rank

Maintainer

Maintained by Kirill Borodin (SpeechAntiSpoofingBenchmarks).

Email: [email protected]
Telegram: @korallll_ai

Fuentes

1 fuente

Hugging FacePrincipalVersión 6d9b482c6f58616253e6be80544912f3c627184a

huggingface.co/SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA/tree/6d9b482c6f58616253e6be80544912f3c627184a

Verificado 25 oct

Download files

Artefactos del modelo

2 artefactos

mms-300m-antideepfake-nda.onnx

onnx · 1,18 GB · SHA-256 291e831c9145…3498 · Hugging Face

Download file

model.safetensors

safetensors · 1,18 GB · SHA-256 ce640c764f41…c8d2 · Hugging Face

Download file

Extractos de fuentes

3 extractos

Readme7 agohuggingface.co/SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA

--- license: cc-by-nc-sa-4.0 base_model: - nii-yamagishilab/mms-300m-anti-deepfake-nda tags: - anti-spoofing - audio-deepfake-detection - speech pipeline_tag: audio-classification --- # MMS-300M-AntiDeepfake-NDA Bit-exact copy of [`nii-yamagishilab/mms-300m-anti-deepfake-nda`](https://huggingface.co/nii-yamagishilab/mms-300m-anti-deepfake-nda) (the AntiDeepfake collection, nii-yamagishilab), hosted for the Speech Anti-Spoofing **Arena**. Trained **without** data augmentation (NDA). - **Paper:** Post-training for Deepfake Speech Detection, [arXiv:2506.21090](https://arxiv.org/abs/2506.21090) - **Original code:** https://github.com/nii-yamagishilab/AntiDeepfake - **Backbone:** MMS-300M; head = AdaptiveAvgPool1d + Linear(D, 2). Score = real logit. - **License:** CC BY-NC-SA 4.0 (non-commercial) — inherited from the source. ## ⚠️ Evaluation restriction It is **STRICTLY PROHIBITED** to evaluate this model on its training / in-domain corpora: ASVspoof2019 LA, ASVspoof2021 LA/DF, ASVspoof5, CFAD, DECRO, DFADD, "Diffuse or Confuse", DiffSSD, DSD, HABLA, MLAAD, SpoofCeleb, CVoiceFake, any LibriTTS-based set (incl. SONAR, CD-ADD, LibriSeVoc), any VoxCeleb-based set, WaveFake, any FLEURS-based set, CodecFake (any variant), any AIShell3-based set, and any MLS-based set. On the Arena it is evaluated **only** on ArAD and In-the-Wild. All credit to the original authors (Ge, Wang, et al., NII). ## Arena results  [![EER% 0.83 on XMAD](https://img.shields.io/badge/EER%25%20on%20XMAD-0.83%25-lightgrey)](https://huggingface.co/spaces/SpeechAntiSpoofingBenchmarks/SpeechAntiSpoofingArena?system=mms-300m-antideepfake-nda) [![EER% 0.64 on PyAra](https://img.shields.io/badge/EER%25%20on%20PyAra-0.64%25-lightgrey)](https://huggingface.co/spaces/SpeechAntiSpoofingBenchmarks/SpeechAntiSpoofingArena?system=mms-300m-antideepfake-nda) [![EER% 5.63 on ODSS](https://img.shields.io/badge/EER%25%20on%20ODSS-5.63%25-lightgrey)](https://huggingface.co/spaces/SpeechAntiSpoofingBenchmarks/SpeechAntiSpoofingArena?system=mms-300m-antideepfake-nda) [![EER% 0.25 on J-SPAW_LA](https://img.shields.io/badge/EER%25%20on%20J--SPAW_LA-0.25%25-lightgrey)](https://huggingface.co/spaces/SpeechAntiSpoofingBenchmarks/SpeechAntiSpoofingArena?system=mms-300m-antideepfake-nda) [![EER% 5.73 on DeepVoice](https://img.shields.io/badge/EER%25%20on%20DeepVoice-5.73%25-lightgrey)](https:/...

Note25 octhuggingface.co/SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA

Source context: 21 downloads · 0 likes · Pipeline audio-classification · Repo SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA

Note7 agohuggingface.co/SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA

Source context: 5 downloads · 0 likes · Pipeline audio-classification · Repo SpeechAntiSpoofingBenchmarks/MMS-300M-AntiDeepfake-NDA