Yan

Yan is a hackathon prototype for practicing real-world English conversations in high-stress newcomer scenarios like airports, shopping, and daily interactions.

The frontend lets users choose a scenario, generate guided dialogue with phonetic hints, and practice the conversation in an interactive scene.

Demo

Technology Stack

AI & Language Models

• GPT-4o - scenario generation and conversational responses
• GPT-5.4 - advanced dialogue generation and reasoning
• Ollama - local model hosting and inference

Speech & Linguistic Resources

• CMU Pronouncing Dictionary (CMUDict) - phonetic transcription and pronunciation support

Backend

• Flask - API services and application logic

Backend code is maintained on the backend branch.

Frontend

• React
• TypeScript
• Vite
• ESLint
• CSS

Local Setup

npm install
npm run dev

Create .env.local with:

VITE_API_BASE_URL=/api/v1
API_PROXY_TARGET=http://127.0.0.1:8000

Start the backend server separately (see backend branch for instructions).

Scripts

• npm run dev
• npm run build
• npm run lint
• npm run preview