A fully open-source AI powered virtual companion that combines chatbot intelligence, voice interaction, and visual avatar movement. Designed to run locally with minimal GPU requirements.
To build a lightweight AI chatbot with a virtual avatar that:
- Runs on your local machine
- Requires minimal hardware resources
- Supports multiple AI backends (local or online)
- Provides both chat and voice-based interaction
- Install ffmpeg on your system.
- Install VTube Studio.
- Download the latest release
.zip. - Extract the archive and rename
Example env.txt→.env. - Launch VTube Studio and enable API Access from settings.
- Run
start.bat. - Grant the plugin permission in VTube Studio.
- Use CTRL + Mic in the terminal to activate speech input.
- Integration with Oobabooga
- Support for Character.ai (beta)
- Local or cloud-based LLMs (via Colab or on-device)
- Real-time avatar interaction with speech & chat
- Windows 7 or newer
- Python
- C++ Compiler (MinGW)
- Git
- Visual C++ Redistributable
- VTube Studio
- API keys for Character.AI & ElevenLabs
- FFmpeg installation help
- For errors, check the Troubleshooting Guide before submitting an issue.
- Voicevox integration
- Oobabooga text generation
- Local & Colab LLM inference
- Local-only execution without Character.AI
- Internet-enabled assistant
- Continuous improvements planned
We welcome community contributions!
- Report bugs via Issues
- Submit pull requests for fixes & improvements
- Propose new features in the discussion board
If you find this project useful, consider supporting via PayPal.
This project builds upon the work of many open-source initiatives:
Special thanks to Alexander Lewis for community contributions.
Licensed under MIT. Copyright © 2023
Clean. Modular. Open-source. Your AI Avatar Companion.