NokiMo
Innovate Futures @ Benji
Innovate Futures @ Benji

patreon


Sonic In ComfyUI - A Truly Awesome Portrait Animation AI Talking Avatar Framework

Tutorial Video: https://youtu.be/8DTAAOhRx3Q

Discover Sonic AI, an open-source framework that transforms static images into animated avatars using audio input—no subscriptions needed! In this tutorial, learn how to install and use Sonic AI in ComfyUI, leveraging Stable Video Diffusion (SVD) for smooth facial animations, lip-syncing, and head movements. Perfect for content creators and AI enthusiasts, this tool rivals paid services like Hydra and Live Portrait while being fully customizable and free! https://jixiaozhong.github.io/Sonic/

Who is this content suitable for?

AI developers, content creators, freelancers, and tech enthusiasts looking to create professional talking avatars without costly subscriptions. Ideal for ComfyUI users familiar with SVD or those exploring AI-driven video generation.

Why does it matter?

Sonic AI democratizes high-quality avatar animation, offering studio-grade results with minimal setup. Its audio-driven workflow, open-source flexibility, and compatibility with tools like Whisper Tiny and SVD XT 1.1 make it a game-changer for video production, presentations, and social media content. Stay ahead of the curve and cut costs with this cutting-edge AI solution!

ComfyUI Setup Steps: 

https://github.com/smthemex/ComfyUI_Sonic/tree/main

Step 1

--------------------------------------------------

git clone https://github.com/smthemex/ComfyUI_Sonic.git

pip install -r requirements.txt

Step 2

--------------------------------------------------

https://drive.google.com/drive/folders/1oe8VTPUy0-MHHW2a_NJ1F8xL-0VN5G7W

https://huggingface.co/openai/whisper-tiny/tree/main

In : ComfyUI/models/sonic/

    |-- audio2bucket.pth

    |-- audio2token.pth

    |-- unet.pth

    |-- yoloface_v5m.pt

    |-- whisper-tiny/

        |--config.json

        |--model.safetensors

        |--preprocessor_config.json

    |-- RIFE/

        |--flownet.pkl

Step 3

--------------------------------------------------

https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt-1-1/tree/main

 ComfyUI/models/checkpoints

    ├── svd_xt.safetensors or svd_xt_1_1.safetensors

Attached the basic workflow used in this tutorial:

Sonic In ComfyUI - A Truly Awesome Portrait Animation AI Talking Avatar Framework

Comments

SONICTLoader 'NoneType' object has no attribute 'seek'. You can only torch.load from a file that is seekable. Please pre-load the data into a buffer like io.BytesIO and try to load from it instead. Anyone running into this error?

Coby Christiansen

looks like you haven't check out the video, and even this post have mentioned.

Benjamin Law

ImageOnlyCheckpointLoader Model in folder 'checkpoints' with filename 'svd_xt_1_1.safetensors' not found.

Coby Christiansen


Related Creators