AI-powered classroom platform for Tunisian Arabic. Fine-tuned Whisper STT for accurate transcription, automated audio processing with Python/Flask, and full-stack app using React, TypeScript, and Node.js. Includes AI Q&A, summarization, and scalable background processing with Redis and MongoDB.
# AI-Powered Classroom Platform for Tunisian Arabic
## 📌 Description
This project is an AI-powered classroom platform designed to enhance learning through automatic transcription, translation, and interactive question answering. The system is optimized for Tunisian Arabic dialect by fine-tuning the Whisper Speech-to-Text (STT) model and building a complete full-stack learning environment.
The platform allows teachers to create sessions, upload audio content, and manage course materials, while students can join sessions, interact, and ask questions. Audio recordings are automatically processed, transcribed, translated, and structured for efficient search and retrieval.
## 🚀 Features
- Fine-tuned Whisper STT model for Tunisian Arabic dialect
- Teacher classroom session creation and management
- Student interaction and course material access
- Automatic audio transcription and translation
- AI-powered Q&A and summarization from course transcripts
- Background audio processing and task queue system
- Structured data storage optimized for search and retrieval
## 🛠️ Tech Stack
### Frontend
- React
- TypeScript
- Tailwind CSS
### Backend
- Node.js
- Python / Flask (Audio processing workers)
### AI & Processing
- Whisper (Fine-tuned)
- LangChain
- ChatGPT API
### Database & Infrastructure
- MongoDB
- Redis (Task Queue)
## ⚙️ How It Works
1. Teachers upload audio or create classroom sessions
2. Audio files are processed using background workers
3. Whisper model transcribes Tunisian Arabic audio
4. Transcripts are translated and chunked
5. Data is stored in MongoDB for fast retrieval
6. Students interact with AI to ask questions and get summaries
## 🎯 Use Cases
- Online classrooms
- Lecture transcription
- AI-powered learning platforms
- Tunisian Arabic speech recognition applications