
(Note: This banner represents the Quantum Adventure theme)
"Empowering Arabic-speaking children to find their voice through Artificial Intelligence & Quantum Mechanics."
AJ is a revolutionary gamified speech therapy platform designed specifically for Arabic-speaking children. By combining OpenAI's Whisper Model for real-time pronunciation analysis with a Quantum Computing Backend (Qiskit) for truly random reward generation, we transform tedious therapy sessions into an exciting treasure hunt.
- The Problem & Solution
- Key Features
- Quantum Intelligence ⚛️
- AI Speech Engine 🤖
- Tech Stack
- Installation & Setup
- Project Structure
- The Team
- Accessibility: Speech therapy in Palestine is expensive with long waiting lists.
- Engagement: Traditional repetition exercises are boring for children.
- Language Gap: Most AI speech tools do not support specific Arabic phonemes (like 'Qaf', 'Ra', 'Sin').
AJ acts as a pocket-sized AI therapist. It uses gamification to keep children engaged and advanced speech recognition to provide instant, gentle feedback. We introduce Quantum Computing to the mix to teach children about science while ensuring fair, non-deterministic rewards.
- Dual Tutors: Children choose their guide—Aws or Joud—creating an emotional connection.
- Story Mode: A treasure hunt where unlocking the chest requires correct pronunciation.
- Dynamic Learning: The app remembers which letters the child struggles with (e.g., swapping 'K' for 'T').
- Targeted Training: Automatically unlocks specific exercises for diagnosed issues (e.g., "The Letter K Module").
- A reward system that exists in Superposition.
- Opening the chest "collapses" the quantum state to reveal either a Legendary or Common prize based on true quantum randomness.
Unlike standard apps that use pseudo-random numbers (algorithms), AJ uses True Randomness generated by a Quantum Circuit.
How it works:
-
Superposition: We initialize a Qubit and apply a Hadamard Gate (H). This puts the prize in a state of being both "Legendary" and "Common" simultaneously
($|0\rangle + |1\rangle$ ). - Measurement: When the child clicks "Open", we measure the Qubit.
- Collapse: The wave function collapses to a single state (0 or 1), determining the prize.
This feature runs on a local Python Flask server using Qiskit.
We utilize OpenAI Whisper (Large-v3) via Server Actions to analyze audio.
- Recording: The child's voice is recorded in high-quality WebM format.
- Processing: The audio is sent securely to OpenAI.
- Comparison: We compare the transcribed text against the target phonemes using Levenshtein distance logic to detect subtle pronunciation errors in Arabic.
- Framework: Next.js 14 (App Router)
- Language: TypeScript
- Styling: Tailwind CSS + Shadcn UI
- Animation: Framer Motion & Canvas Confetti
- BaaS: Supabase (PostgreSQL)
- Auth: Supabase Magic Links
- AI Model: OpenAI Whisper API
- Quantum SDK: Qiskit (IBM Quantum)
- Quantum Server: Python 3.10 + Flask
This project uses a Hybrid Architecture (Node.js Frontend + Python Backend).
git clone https://github.com/awshammad04/aj.git
cd ajnpm install
cp .env.example .env.local
npm run devFrontend runs on: http://localhost:3000
cd quantum-backend
pip install flask flask-cors qiskit qiskit-aer
python app.pyQuantum Server runs on: http://localhost:5000
AJ/
├── app/
│ ├── assessment/
│ ├── training/
│ └── quantum-chest/
├── components/
├── lib/
├── quantum-backend/
│ └── app.py
└── public/Built with ❤️ for the AI & Quantum Computing Hackathon 2025/2026.
- Aws Hammad – Birzeit University
- Joud Rami – Palestine Polytechnic University