# Ayah Lense

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Faraz Syed](https://cerebralvalley.ai/u/khirad_ai)
- **GitHub:** https://github.com/khiradlabs/ayah-lens
- **Demo video:** https://github.com/khiradlabs/ayah-lens/releases/download/v1.0-hackathon/Ayah-Lens-Demo.mp4
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4156

yah Lens brings the meaning of Quran recitation into view on Meta Ray-Ban Display glasses. The listener hears a reciter; the verse and its published English appear on the lens and follow the recitation on their own. Nothing is tapped.

Impact. About two billion Muslims (Pew, 2020 estimate) hear the Quran daily, many without following the Arabic. Looking a verse up on a phone breaks the moment; Ayah Lens keeps the meaning in view, hands free. Today it does one thing well. With more development the same foundation can support those memorizing the Quran, flag a wrong recitation for a learner, and bring the meaning into any language.

Use of Meta AI tooling. The Wearables Device Access Toolkit 1.0.1 connects the glasses’ microphones and display to a native iPhone app. Muse Voice Transcribe (muse-voice-transcribe-1.0, realtime ASR) returns live Arabic. Muse Spark 1.3, through the chat completions API with strict JSON-schema output, resolves look-alike verse candidates within a fixed list or abstains, and pairs Arabic spans with the published Pickthall English; each plan is validated for source identity and coverage, then cached.

Technical depth. A local index covers all 114 surahs and 6,236 ayahs; a source-word tracker follows the listener’s position inside a single verse, shown live on 2:282, the longest in the Quran. The lens shows only unchanged Tanzil Arabic and Pickthall English. 59 tests cover matching, source preservation, portions and scheduling. Transcription and tracking run live; only Spark’s English pairing is prepared in advance.

Demo. Real glasses display recordings: a verse arriving and advancing by itself, the portion moving inside one long verse, the founder on camera. Software written in ten days, Oct 26 to Nov 4. Code: github.com/khiradlabs/ayah-lens.

## More from Meta Global AI Developer Hackathon

- [TempoAI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4153)
- [OptiPack](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4154)
- [EmberWalk](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4155)
- [Unwritten](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4157)
- [THE CODE-BREAKERS](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4158)
- [Riris Nurul Aini](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4159)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4156. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt