Spaces:

amphion

/

DeepfakeDetection

like 3

Running on Zero

Dear Hugging Face Team,

I'm writing to apply for the GPU Community Grant on behalf of our academic project, Amphion, which is being developed at The Chinese University of Hong Kong, Shenzhen.

Our team, composed of devoted doctoral students, graduate students, and undergraduate students, is enthusiastic about our work and dedicated to making contributions to the academic and open-source community.

In this project, we are focused on constructing deepfake detection models and contributing to the field of AI security. Specifically, the biggest issue with audio deepfake detection currently is the problem of generalization. In our research, we adopt the idea of in-context learning. When identifying whether a test sample is generated by AI, we first provide the model with several sample prompts. These prompts enable the model to concentrate more on similar fake information and improve its generalization ability.

We would be extremely grateful for the support of the GPU Community Grant to further our research and development endeavors. With this support, we can continue to innovate and contribute to the technological and academic community.

Thank you for considering our application. We're looking forward to the possibility of collaborating with Hugging Face and having a significant impact on the field of audio deepfake detection.

Best Regards,
Li Wang