Welcome to my 💫🎙️ Friendly InfiniteTalk Vibe!

✨ Less mess, more magic.

InfiniteTalk and VibeVoice in one package. Now you don't have to strain yourself generating voices and lipsync using third-party resources or workflows. InfiniteTalk and VibeVoice are made for each other, as both models understand emotions and also allow you to generate long videos (up to 45 minutes on the VibeVoice Large model).

🚀 Don't forget to update ComfyUI and nodes in the workflow for the best perfomance!

💻 System requirements for 480p resolution:

• Minimum system requirements:

RTX 3000-s, 8-10 GB of video memory, 45 GB of RAM, 8-core processor, SSD, latest ComfyUI

• Recommended system requirements:

RTX 4000-s or higher, 16 GB of video memory, 64 GB of RAM, 8-core processor, PCI-E 4.0 SSD, latest ComfyUI

📌 Detailed tips in the workflow

✨ Workflow features:

• Friendly UI without spaghetti connections

• Convenient step-by-step generation, where you first generate the desired voice and then move on to video generation

• Use GGUF or regular models (links in the workflow)

• Ability to generate a voice or use existing audio

• Fine-tune voice generation parameters

• Convenient settings slider control

• Output volume normalization has been configured, as VibeVoice initially generates quiet audio

• Optimization and memory reduction between stages

• Settings, detailed tips, and links to models

• Interpolation module for the smooth video

🤗🙏🏼 Thanks to MeiGen-AI and kijai

Original repo — GitHub