AI-Waifu-Vtuber

Why we included this project

Streamers who want an always-on AI sidekick on camera will get the most from this project: it turns a large language model into a voiced character that listens, replies in character, and reads and answers Twitch chat out loud. Under the hood it connects OpenAI Whisper for speech-to-text, your choice of VoiceVox or Silero TTS for the voice, and DeepL for translating incoming messages, then sends the result to VtubeStudio so the model lip-syncs as it talks. The character's personality lives in simple text files, so swapping in a new persona is mostly editing a few lines instead of touching code. Setup expects you to bring your own OpenAI API key, TTS engine, and a virtual audio cable, so it's aimed at tinkerers comfortable hooking up several parts rather than people wanting a one-click install. It also works as a reference if you want to build your own talking avatar or voice-controlled stream bot.