I did this custom nodes for ComfyUI that implementing the CSM model for text-to-speech generation
Hope someone finds it useful
Link to the nodes https://github.com/thezveroboy/ComfyUI-CSM-Nodes
Description
Comments (2)
Thanks, I'll try it out. Sucks it apparently needs Torch 2.4.0 though.
You can remove the version requirements from the requirements.txt file. A huge limitation of this node is that you can't provide it with context audio clips, so the voices just end up being random.