Spaces:
Running
on
Zero
Apply for a GPU community grant: Academic project
Project: SoulX-Singer — Zero-Shot Singing Voice Synthesis
SoulX-Singer is an open-source singing voice synthesis system capable of generating natural singing audio from lyrics and melody (MIDI) with zero-shot voice generalization.
We are releasing:
• model weights
• inference pipeline
• web demo
• documentation
The Space will allow researchers, musicians, and creators to:
- synthesize singing from lyrics + MIDI
- experiment with controllable vocal expression
- study singing synthesis and voice cloning
The ZeroGPU resource will only be used for inference in the public demo Space. We optimize generation latency and queue requests to avoid resource abuse.
This demo provides a rare open and reproducible SVS system to the community, since most current singing synthesis systems are closed or require complex setup.
Repository:
https://huggingface.co/Soul-AILab/SoulX-Singer
https://github.com/Soul-AILab/SoulX-Singer
Technical Report
Hi
@Xinsheng-Wang
, we've assigned ZeroGPU to this Space. Please check the compatibility and usage sections of this page so your Space can run on ZeroGPU.
If you can, we ask that you upgrade to Enterprise to enjoy higher ZeroGPU quota and other features like Dev Mode, Private Storage, and more: hf.co/enterprise
Hi @Xinsheng-Wang , due to some dependencies and nemotron, there are some challenges running it with ZeroGPU as you may have noticed. I manged to fix it and streamline the UI a bit, let me know if you like it, if so I can PR it here: https://huggingface.co/spaces/multimodalart/SoulX-Singer