Spaces:

---

Duplicated from  HumeAI/tada

fffiloni

/

tada-dual-alignment-tts-demo

like 22

Running on Zero

Hi! 🤗

I’ve been working on an improved demo of the TADA project and was wondering if it would be possible to enable ZeroGPU for the Space.

The goal isn’t just to mirror the original demo — I rebuilt it with a clearer UI, a more guided workflow, and some pedagogical elements to better highlight the Text-Acoustic Dual Alignment idea from the paper.

I also made several technical improvements to make it behave better on Hugging Face Spaces, including:

• proper HF OAuth login for gated models
• memory-safe model loading and GPU offloading
• compatibility with both fixed GPU and ZeroGPU environments

Would it be possible to enable ZeroGPU for this Space?

Thanks a lot!