This project provides an AI-powered interview preparation tool, enabling users to practice interviews with real-time speech-to-text transcription and AI-generated answer suggestions. It targets individuals preparing for technical or behavioral interviews, offering a comprehensive solution for honing responses and managing practice sessions.

How It Works

The application integrates real-time audio processing via the WebAudio API with iFlytek's RTasr API for high-accuracy speech-to-text transcription. Transcribed speech is then fed to DeepSeek AI, which generates intelligent answer suggestions. The backend, built with FastAPI, handles user authentication (JWT), manages interview sessions, and persists data using PostgreSQL and Redis. Frontend components are built with React 19, TypeScript, and Vite, utilizing Ant Design for the UI and Zustand for state management, ensuring a responsive and interactive user experience with real-time data synchronization.

Quick Start & Requirements

• Primary install/run command: Docker is recommended. Use scripts/docker-start.sh for production or scripts/dev-start.sh for development. docker-compose commands are also provided.
• Non-default prerequisites: Node.js 16+, Python 3.9+, PostgreSQL 17+, Redis 7+. API keys for DeepSeek AI and iFlytek are required and configured within the application.
• Links:
  • iFlytek API: https://www.xfyun.cn/
  • DeepSeek AI API: https://platform.deepseek.com/
  • API Docs: http://localhost:9000/docs (Swagger UI), http://localhost:9000/redoc

Highlighted Details

• Real-time STT powered by iFlytek RTasr API.
• AI-driven answer suggestions leveraging DeepSeek AI.
• Full interview session management with history and real-time synchronization.
• Secure user authentication via JWT.
• Modern tech stack: React 19/TypeScript/Vite (frontend), FastAPI/PostgreSQL/Redis (backend).

Maintenance & Community

The project roadmap outlines development milestones up to v0.5. Contribution is encouraged via pull requests, with WeChat contact provided for direct engagement with the maintainer, "JarvanJason".

Licensing & Compatibility

The project is released under the MIT License, permitting commercial use and integration into closed-source projects.

Limitations & Caveats

The project description notes that "difficulty controlling the timing of AI questions, low accuracy, needs improvement" is an area the project aims to address, suggesting potential limitations in the current AI interaction logic or STT accuracy under certain conditions. Configuration of external API keys is a mandatory setup step.