About
Welcome to the dedicated cloud repository for Wunjo AI's neural network models. Unlock the unparalleled capabilities of neural networks with Wunjo AI. Whether you're delving into speech synthesis, crafting deepfake animations, drawing Stable Diffusion video by text prompt or video making, Wunjo AI has got you covered.
In version 1.6, Wunjo supports multiple languages for voice cloning and speech synthesis, TTS model training, real-time speech recognition, audio enhancement, a multi-language interface, and the use of the "pass" keyword in the video-to-video feature.
The new version 2.0 introduces a completely redesigned interface, enhanced performance, improved face swap functionality, a generator for non-existent people’s faces, and a deepfake analyzer. It also includes improvements in lip sync, content enhancement, object, text, and background removal, as well as content and component restyling using text prompts. Additionally, public links enable community creation and collaboration.
Future updates will reintroduce and enhance the missing audio functionalities from version 1.6, making them even better.
Available as both a free Community Edition and a subscription-based Professional Edition on wunjo.online, Wunjo caters to a wide range of users, from beginners to professionals, ensuring privacy by operating locally on your device.
Why Choose Wunjo Community Edition?
- All-in-One: A comprehensive tool catering to both your voice and visual AI needs.
- User-friendly: Designed for all, from beginners to professionals.
- Privacy First: Functions locally on your desktop, ensuring your data remains private.
- Open-source & Free: Benefit from community-driven enhancements and enjoy the app without any cost.
Why Choose Wunjo Professional?
- More features: First access to updates, additional and more advanced features.
- Public links: Host your own server (PC) application accessible from anywhere in the world.
- Unlimited performance: Process multiple tasks simultaneously to increase overall productivity.
Step into the future of AI-powered creativity with Wunjo.
Support the Project
You can support the author of the project in the development of his creative ideas on Boosty or CloudTips. Thanks to Boosty, you will have the opportunity to use your voice to choose which functions from the Professional version will appear in the Community Edition in new updates. There are other ways to support the development of the project, more details on page.
Credits
Wunjo is built on the work of many open-source projects. This repository keeps
copies of the weights its plugins download, so the plugins keep working if an
original repository moves. Every file is an unmodified copy unless it is marked
as trained by me, and every folder keeps the licence of its source: each
subfolder holds its own LICENSE and a README.md naming the source and the
commit it was copied from. Several of these licences forbid commercial use.
Check the folder before you use a file outside Wunjo.
Models used by the Wunjo Make plugins
| Folder | Plugin | Project | Licence |
|---|---|---|---|
clone/converter.pth, clone/config.json | Voice Toolkit | OpenVoice tone colour converter | MIT |
clone/clone-0.6b/, clone/clone-1.7b/, clone/design-1.7b/, clone/speech_tokenizer/ | Voice Toolkit | Qwen3-TTS by the Qwen team, Alibaba Cloud | Apache-2.0 |
separator/mdx/ | Voice Toolkit | MDX-Net models from Ultimate Vocal Remover | MIT |
lips/latentsync/ | Lip Sync | LatentSync by ByteDance | CreativeML OpenRAIL-M |
lips/whisper/ | Lip Sync | Whisper tiny by OpenAI | MIT |
lips/sd-vae-ft-mse/ | Lip Sync | sd-vae-ft-mse by Stability AI | MIT |
lips/fan/ | Lip Sync | FAN from face-alignment by Adrian Bulat | BSD-3-Clause |
deepfake/liveportrait/ | Face Toolkit | LivePortrait by Kuaishou | MIT |
deepfake/faceswap.onnx | Face Toolkit | face swap model from InsightFace | non-commercial research only |
depth-map/small/ | Depth Map | Depth Anything V2 Small | Apache-2.0 |
depth-map/base/, depth-map/large/ | Depth Map | Depth Anything V2 Base and Large | CC BY-NC 4.0 |
foley/mmaudio/ | Foley | MMAudio networks and VAE | CC BY-NC 4.0 |
foley/mmaudio/synchformer_state_dict.pth | Foley | Synchformer by Vladimir Iashin | MIT |
foley/bigvgan/ | Foley | BigVGAN v2 by NVIDIA | MIT |
foley/clip/ | Foley | DFN5B-CLIP-ViT-H-14-384 by Apple | Apple ML Research licence, research only |
qwen35/ | Qwen 3.5 | Qwen3.5 4B and 9B, GGUF by Unsloth | Apache-2.0 |
The face recognition and detection models of InsightFace that Face Toolkit also uses are not copied here; the plugin downloads them from their own source.
Models of earlier Wunjo versions
Kept for the older desktop releases that still download them.
| Folder | Project |
|---|---|
rtvc/ | Real-Time Voice Cloning, trained by me for English, Russian and Chinese; rtvc/general/signature.pt is my own audio watermark model |
deepfake/emo2lip.pth, lips/synclip_v2.model | lip sync built on Wav2Lip |
deepfake/vit_b_quantized.onnx, deepfake/vit_h_quantized.onnx | Segment Anything, quantized |
deepfake/vgg16_*.pth, deepfake/retouch_*.pth | text detection and retouching models of Wunjo 2.0 |
ebsynth/ | EbSynth |
separator/mdx/MDX_Inst_Main.onnx, separator/mdx/MDX_KARA_2.onnx | MDX-Net models from Ultimate Vocal Remover |
I extend my deepest gratitude to the authors of these projects. If you would like to contribute a model, for example voice cloning for a new language, reach out through GitHub Discussions or the Hugging Face discussions.