Wunjo CE is an open-source, offline, and free AI toolkit for visual and audio content manipulation. It targets users from beginners to professionals, offering features like face swapping, lip-syncing, object removal, voice cloning, and video generation, all processed locally for privacy.

How It Works

Wunjo CE leverages a suite of adapted open-source AI models, including Wav2lip, Insightface, Stable Diffusion, and ControlNet, for its diverse functionalities. It processes tasks locally, supporting both CPU and GPU (Nvidia via CUDA, AMD via ZLUDA) for enhanced performance. The project emphasizes a user-friendly interface and offers an API for developer integration.

Quick Start & Requirements

• Installation: Official installers available for Windows/Ubuntu, or via Docker. GitHub Actions can build installers.
• Prerequisites: Python 3.10, ffmpeg. Nvidia GPU with CUDA, or AMD GPU with ZLUDA. firmware-linux-nonfree package required for AMD GPU availability.
• Setup: Detailed instructions in the GitHub Wiki.
• Links: Official Website, GitHub Repository, Changelog.

Highlighted Details

• Animate Portrait Mode and Retarget Portrait for copying facial expressions.
• Neural network for automatic video summarization and content analysis.
• Control restyling with 8GB VRAM, allowing changes to objects, gender, and nationality.
• Voice cloning in any language from text and audio, with improved audio separation.
• Deepfake analyzer to distinguish authentic from manipulated media.

Maintenance & Community

The project is actively developed by Wladislav Radchenko, with community support encouraged via GitHub stars and contributions. Discussions and feature requests are managed on GitHub. Contact via email: i@wladradchenko.ru.

Licensing & Compatibility

The project is open-source. Specific license details are not explicitly stated in the README, but the emphasis on "Open Source, Local & Free" suggests a permissive license. Commercial use is possible via the "Wunjo Pro" subscription.

Limitations & Caveats

AMD GPU support relies on the ZLUDA project, which may have its own compatibility or stability limitations. Some advanced features are exclusive to the paid "Wunjo Pro" version. The project aims for 4096 GitHub stars to unlock further updates for the Community Edition.