LFQIH BOT

Inspiration

The idea for Lfqih Bot came from a deeply personal moment. After moving away from home, my mother once said: "Since you left, I haven't had anyone to help me review the Qur’an like before." This touched me. In Morocco, it's common to recite and review the Qur’an with a partner — a family member, a teacher, or a fqih. But not everyone has someone to help them. That’s when I thought: “What if I could build an AI that listens like a fqih, corrects mistakes, and motivates you — all through your voice?” This project is for everyone who wants to memorize the Qur’an but doesn’t always have someone to guide them.

What it does

Lfqih Bot is an AI-powered assistant that helps users memorize the Qur’an through voice interaction. Here’s what it can do: You tell it which Surah and Hizb you want to review. It listens to your recitation in Arabic. If you make a mistake or pause, it corrects you gently and gives you the next word or verse. If you complete the passage without mistakes, it congratulates you warmly . The experience is meant to feel like reciting with a real Moroccan fqih or a caring family member.

How we built it

Speech Recognition (ASR): We used Arabic-specific ASR models OpenAI Whisper fine-tuned for Qur’anic pronunciation. Qur’an Dataset: Integrated a structured database of the Qur’an, broken down by Surah, Ayah, and Hizb for real-time tracking. Error Detection: Compared recited audio to expected Ayat to detect missing, incorrect, or skipped words. Voice Interaction: Built a lightweight voice-controlled interface using Python and Flask.

Challenges we ran into

Arabic ASR accuracy: Arabic dialects and tajweed rules made accurate recognition a big challenge. Recitation Variations: Users pronounce differently (pauses, elongations), and aligning that with exact Ayat required tolerance logic. Real-time corrections: Building a system that interrupts and guides without frustrating the user took UX testing. Offline support (optional): We wanted it to be usable without constant internet, which meant working with local models or compressed deployments.

Accomplishments that we're proud of

Built an Arabic voice assistant that can recognize Qur’anic recitation — a rare and complex use case. Created an experience that’s deeply personal and culturally relevant to millions of users. The project made our team, especially our families, proud — and we got emotional user reactions from our test demos.

What we learned

How to build real-time voice interaction apps for Arabic, a language often underrepresented in AI tools. How to make AI feel human, warm, and supportive — not just functional.

What's next for LFQIH BOT

Add tajweed mistake detection (not just missing words). Support multiple Arabic dialects and recitation styles (قراءات). Release on mobile platforms (Android & iOS) with voice-first UX.