Holy Quran Speech Recognition System (HQSR)
A research-driven Android application for Holy Quran speech recognition that achieved a 97.6% F1 score, combining robust ML sequence modeling with production-ready API architecture for practical recitation analysis.
Business Challenge
The project required highly reliable Quran recitation recognition for real users while handling noisy, diverse audio conditions and preserving model consistency across mobile-first usage scenarios.
Our Solution
Built an Android app using Flutter and Supabase, integrated FastAPI for model serving, and engineered an LSTM-based speech recognition pipeline with domain-specific preprocessing techniques that delivered a validated 97.6% F1 score.
97.6% F1 Accuracy Benchmark
Demonstrated high-confidence AI performance with a 97.6% F1 score, validating model quality for practical Quran recitation recognition.
Mobile-First Accessibility
Delivered the system through an Android Flutter app to make advanced recognition features accessible to students and learners.
Research-to-Product Execution
Combined LSTM experimentation, API integration, and production-friendly architecture into a practical end-to-end prototype.
Project Screens
Client
Research Project
Category
AI,Mobile
Year
2026