🕌 Mishkat: Multisensory Quranic Rendering Engine

Mishkat is the "Global Reach" arm of the Bayan ecosystem. It serves as a technical Proof of Concept (PoC) for transforming semantic understanding into Context-Aware Audio and visual experiences. It represents a pioneering model in Disciplined Generative AI, designed to bridge the emotional gap for non-Arabic speaking children.

---

📖 The Vision: Addressing the "Emotional Gap"

1. The Motivation: Bridging the Language Barrier via Intuition

Mishkat starts from a global civilizational challenge: How do we connect a non-Arabic speaking child to the Quran before they master the language?

We believe the Quran is not merely text to be read, but an Experience. Mishkat aims to address the "auditory and visual intuition" of the child. Our goal is not just to teach recitation, but to build an early emotional bond, making the child feel the Quran is addressing them in their mother tongue, with a familiar tone, transforming linguistic alienation into emotional familiarity.

2. The Technical Gap: Taming "Blind" Generative AI

Despite the GenAI revolution, current applications suffer from "Cognitive Disconnect":

• The Problem: Current Text-to-Video and TTS tools act as "black boxes," isolated from the sanctity and context of the Quranic text, often producing random or spiritually inappropriate content.
• The Challenge: How do we tame these wild models to build a Controlled Pipeline that generates audio-visual content derived from a deep semantic understanding of the Ayah, rather than superficial generation?

---

🛠️ System Architecture & Innovation

Mishkat introduces the first engineering framework for Multisensory Quranic Rendering.

🧩 The Pipeline Components

1. Emotional Voice Cloning (via Ilqa): We utilize our sister project, Ilqa, to handle the audio generation pipeline.

   • Innovation: Developing TTS models that simulate a "Child Persona".
   • Impact: Presenting Quranic translations in the child's native language (English, French, etc.) using a voice they identify with, creating immediate psychological resonance.

2. Scalable Production Pipeline: To prove the infrastructure's robustness, we have moved beyond sampling. We have successfully automated the generation of over 140,000 video clips covering full Qurans in 20 global languages (out of 83 targeted). This demonstrates immediate Global Scalability.

---

🎮 Interactive Demo & Usage

We provide a Google Colab Notebook as a prototype to demonstrate the generation capabilities.

🚀 Launch Mishkat Demo in Colab

1. How to Use

Select the Surah, Ayah, Reciter, Language, and Visual Template to generate a video instantly.

(Figure 1: Step-by-step guide to generating a video using the Colab interface)

2. Visual Templates

Mishkat supports dynamic visual themes (e.g., "Quran Garden", "Sky & Stars", "Flashcards") to suit different educational contexts.

(Figure 2: Examples of different visual templates generated by the engine)

---

⚖️ Research Impact & Reliability

Mishkat establishes a new standard for AI for Faith-Based Education:

• Meaning-Rendering Separation: The system strictly separates Authoritative Meaning (approved translations) from Sensory Rendering (Audio/Video). The AI never generates the translation text itself; it only renders approved data.
• Controlled Generative Pipeline: All outputs undergo strict semantic constraints. The AI's role is limited to educational representation, not interpretation.
• Precomputed Scalability: Due to the high computational cost of emotional cloning and visual alignment, we adopted a pre-computation strategy, resulting in ~140,000 ready-to-stream assets.
• PoC for Future Systems: This pipeline serves as the foundation for future "Context-Aware Audio" systems capable of coloring audio performance based on meaning (Sadness, Glad Tidings, Warning) under human supervision.

---

🚀 Installation (Local Development)

To run the pipeline locally instead of on Colab:

# 1. Clone the repository
git clone https://github.com/YourOrg/Mishkat.git
cd Mishkat

# 2. Install dependencies
pip install -r requirements.txt

# 3. Install system requirements (Linux/Ubuntu)
sudo apt-get install libreoffice ffmpeg

# 4. Run the pipeline (Example)
python -m src.pipeline.create_ayah_video

---

🤝 Contributing

We welcome contributions, especially in expanding the templates/ library or optimizing the FFmpeg rendering parameters.

1. Fork the repository.
2. Create your feature branch (git checkout -b feature/NewTemplate).
3. Commit your changes.
4. Push to the branch.
5. Open a Pull Request.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

<p align="center"> <strong>Mishkat</strong> part of the <strong>Bayan</strong> Ecosystem.<br> <em>Bridging the gap between Sacred Text and Human Emotion.</em> </p>