Re:View is a mobile application that takes in raw lecture recordings and lecture slides to reconstruct them into a complete easy-to-follow lecture. By combining audio cleanup, accurate transcription, and slide-to-sentence alignment, our application provides realistic lecture experience from the reconstructed lecture presented in a video-like format. Re:View also provides refined lecture transcripts along with automatic annotations for those with hearing disabilities. To summarize, Re:View is a seamless review tool that makes learning easier, more efficient, and more inclusive for everyone involved in education, regardless of their role.
- AI-Powered Lecture Reconstruction: Combines ASR, slide matching, and TTS to create synchronized lecture materials
- Bilingual Support: English and Korean lectures with automatic bidirectional translation
- Intuitive Mobile Experience: Portrait/landscape modes with real-time slide-transcript synchronization
- Cross-Platform Support: Built with Flutter for seamless deployment across Android and iOS
- Accessibility-First Design: Refined transcripts and automatic annotations for hearing-impaired users
- Customizable Interface: Light/dark themes, adjustable playback speed, and TTS settings
- Organized Library: Subject-based organization with tags, favorites, and powerful search
Frontend: Flutter mobile app with Hive-based state management
Backend: FastAPI server with AI inference pipeline (NVIDIA Parakeet, OpenAI Whisper, Hunyuan MT, Kokoro TTS)
cd frontend
flutter pub get
flutter runcd backend
./setup.sh # One-time setup (creates conda environment)
conda activate swpp-backend
python main.py # Server runs on http://localhost:8080For detailed project documentation, design documents, and specifications: