Quran Hifz Coach — recitation verification
FastAPI + Flutter platform for Qur'an memorisation. Audio-to-text verification across three riwāyāt with per-word Tajwīd feedback and adaptive practice plans.
- Client
- Quran Hifz Coach
- Industry
- EdTech
- Year
- 2025
- Duration
- 18 weeks
The problem we walked into.
Verifying recitation in three different Qur'anic styles (Hafs, Warsh, Qalun) with per-word Tajwīd hinting is genuinely hard. The audio pipeline had to be accurate, observable, and i18n-ready for English, Arabic and Bangla — and ship behind TLS with rate limits.
How we framed it.
A FastAPI backend with a DP aligner for audio sync, tokenizer + scorer for verification, a Tajwīd text engine, JWT auth with role gates, and Redis rate limits. Flutter on the front end with audio recorder, RTL UI, and full i18n. Kubernetes ingress with TLS/HSTS, Prometheus scrape, structured JSON logs.
What we shipped.
/v1/verify accepts WAV or URL audio and returns per-word feedback spans. Adaptive coach planner generates practice sessions from each learner's history. Admin dashboard with attempt history, performance metrics, and Prometheus integration. Multi-riwāyah JSON corpus with per-word audio timings.
- Audio verification with DP aligner + per-word Tajwīd scoring and feedback spans
- Three Qur'anic riwāyāt (Hafs, Warsh, Qalun) — full text + per-word audio timings
- Adaptive coach planner generating practice sessions from each learner's progress
- JWT auth with role-based access (learner, admin); Redis rate limiting
- Admin dashboard: learner stats, attempt history, performance metrics, Prometheus
Roktobindu — nonprofit website
A clean, fast Next.js site for a Bangladesh blood-donation nonprofit — mission, project listings, donor flow — delivered in eight weeks.
Have an idea worth building?
Tell us about your project. We'll come back within 24 hours with a plan and a quote.
Start a project