bro77xp/beginner-friendly-ai-vtuber

Beginner-Friendly AI VTuber is a Python project that creates a virtual streamer (VTuber) powered by three open-source AI tools working together. You speak into your microphone, the software transcribes your words, generates a response using a local AI model, speaks the reply aloud in a customizable voice, and at the same time animates a 2D character in VTube Studio to match the speech. The whole loop runs continuously until you stop it. The three core tools are Whisper for speech-to-text (listening and transcribing what you say), Ollama running a local Llama 3.2 model to generate the VTuber's replies, and Chatterbox TTS to convert those replies into spoken audio. The system adapts to your microphone's background noise automatically, so you do not need to configure audio levels by hand. VTube Studio integration lets the character's mouth move in sync with the spoken response. The script also detects emotions in the AI's replies (happy, sad, angry, thinking, neutral) and triggers matching animation hotkeys you configure in VTube Studio. There is also optional Twitch chat integration so the VTuber can read and respond to messages from your stream's chat. Setting up requires Python 3.10, a Python virtual environment, Ollama installed and running, and VTube Studio if you want the animated character. You install the Python dependencies with one pip command, pull the Llama model with one Ollama command, and run the main script. Voice options range from pre-trained Chatterbox voices to custom voice cloning using your own audio recordings. The project runs primarily on CPU, so an AMD or Nvidia graphics card is not required, though processing will be faster with one. No license is stated in the README.