A realistic AI-driven clone of myself, using my face and voice, to chat and interact with.
This is the output of my AI avatar chat project, using Google speech recognition, OpenAI chat completion and Elevenlabs voice cloning.
The code repo is covering the AI-chat interface only (yellow blocks), not the lipsync generation (in green) or the avatar creation and rendering (in red), which are done in 3rd-party software.
Clone the project, create a virtual environment, activate it and install all the modules listed in the requirements file.
This project is using Python 3.12.1 but should work on earlier python3 versions too.
git clone https://github.com/alexdjulin/virtual-ai-avatar.git
cd virtual-ai-avatar
python -m venv .venv
.venv/Scripts/activate.bat
pip install -r requirements.txtRename config_template.yaml in config.yaml and edit or complete the values inside, like your API keys, AI models and settings, avatar info, paths, etc.
Call main.py to start the tool and chose between the Create Avatar Story and the Chat With Avatar modes.
Alternatively you can pass arguments to start in Chat mode directly and specify input and output methods (text/voice and language) if you know the corresponding choice.
# chose mode
python main.py
# input voice de_DE, output voice
python main.py -i 3 -o 2