Talk with a live2d model.

Run locally in the browser or using backend proxy connection Openai, Ollama etc.

Try it online https://live2d-ai-chat.hitorisama.org/

Feature

1. show live2d model 🆗
2. auto change expression of model
3. auto change motion of model 🆗
4. speech to text 🆗 (web speech api)
5. text to speech 🆗 (browser:vits-web; backend: node-edge-tts)
6. style of speech
7. subtitle of AI and User 🆗
8. long-term memory
9. Custom chat model
10. Speaking first / Find topics 🆗
11. changeable model, expression and motion
12. other function: playing games, singing，searching google, etc.

how to develop?

1. install ollama and pull a model you like
2. install nodejs, pnpm, bun(optional)
3. git clone https://github.com/zoollcar/live2d-AI-chat
4. cd live2d-AI-chat & pnpm install & cd backend & pnpm install
5. run the backend: cd backend & cp .env.local.example .env.local & node index.js
6. run the app: cd live2d-AI-chat & pnpm run dev

how to build to exe? (have some issue for now, not work)

1. install nodejs, pnpm, bun
2. git clone https://github.com/zoollcar/live2d-AI-chat
3. cd live2d-AI-chat & pnpm install & cd backend & bun install
4. build the backend embed to frontend : cd backend & bun run build:windows
5. cd live2d-AI-chat & pnpm run tauri:build

config

frontend LLM model: LLMChatWebLLM

frontend TTS model:vitsWeb

backend: .env.local.example

credits

live2d model: Tianyelulu