Memoria

Memoria is a photo-library search app for iOS and Android that indexes what is inside your pictures and videos rather than just their filenames and dates. It runs OCR over stills and video frames, transcribes spoken audio, clusters faces, and recognises objects — then folds all of it into one full-text index you query with an ordinary search box.

The whole pipeline runs locally. There is no account to create and nothing is uploaded; the only network request the app makes is an optional one-time download of the speech model.

What it can find

• Text in images — whiteboards, receipts, contracts, screenshots and memes, including text that only appears in a single video frame. OCR covers seven languages.
• Spoken words in video — recorded lectures, meetings and voice notes are transcribed with a locally executed Whisper model spanning 99+ languages, so searching "derivative" surfaces the lecture clip where it was said.
• People — faces are detected and grouped into clusters on the device, turning "every photo of Emma" into a single tap.
• Objects and metadata — recognised objects sit alongside date, location and camera details in the same query.

Who it is for

Anyone whose camera roll has become an unsearchable archive: students filming lectures, tradespeople photographing invoices and serial plates, journalists working from recordings, and privacy-conscious users who will not hand a photo library to a cloud service.

Practical details

Memoria requires iOS 16.6 or later, and is also published on Google Play. Indexing runs in the background and reports progress, so a large library is processed over time rather than blocking first use. The app is free to download with a usage allowance; Memoria Plus unlocks the full library for a single one-time purchase rather than a subscription. Anonymous usage analytics are opt-out in settings.