EchoMask official checkpoints

Official checkpoints for EchoMask: Speech-Queried Attention-based Mask Modeling for Holistic Co-Speech Motion Generation (ACM Multimedia 2025), a speech-conditioned approach to holistic co-speech gesture generation.

• Paper · Hugging Face Paper page
• Project page · Code
• Published version
• Generated inference data

Files

Archive	Protocol	Contents
weights_echomask_no_smplx.zip	Speaker 2, paper protocol	Original pretrained representation models and EchoMask checkpoint; SMPL-X files excluded
EchoMask_all_speakers_weights_25spk.zip	25 English BEAT2 speakers	Reusable representation models and selected EchoMask checkpoint

The two archives implement different training protocols and should not be treated as a controlled single-speaker versus multi-speaker ablation. See the code repository for installation paths and configuration details.

Verify both downloads with sha256sum -c SHA256SUMS.

Reported results

Protocol	Speakers	FGD ↓	BC ↑	DIV ↑	MSE ↓	LVD ↓
Speaker 2 (paper)	1	0.4623	0.7738	13.370	6.761e-8	7.290e-5
Released all-speaker checkpoint	25	0.5656	0.4951	9.299	4.700e-8	6.090e-5

The first row is from the paper. The second row is from the released all-speaker checkpoint. Their protocols differ.

Terms

This repository mirrors research artifacts released by the authors. No new license is granted by this model card. The Hugging Face archives intentionally exclude SMPLX_NEUTRAL_2020.npz; obtain SMPL-X directly from the official source under its terms. Code, BEAT2, pretrained encoders, and other third-party assets remain subject to their respective terms.

Citation

@inproceedings{zhang2025echomask,
  title={EchoMask: Speech-Queried Attention-based Mask Modeling for Holistic Co-Speech Motion Generation},
  author={Zhang, Xiangyue and Li, Jianfang and Zhang, Jiaxu and Ren, Jianqiang and Bo, Liefeng and Tu, Zhigang},
  booktitle={Proceedings of the 33rd ACM International Conference on Multimedia},
  pages={10827--10836},
  year={2025},
  doi={10.1145/3746027.3754847}
}