How OCR works: from photo to translation in a second
Three of our apps — Photo Translator, Language Translator and PDF Scanner — are built around one technology: optical character recognition (OCR). Here's how it works, in plain words.
Step 1. Camera and frame preparation
Before recognizing text, the frame needs preparation: fixing perspective, raising contrast, removing noise. For a document scanner this is crucial — it's why a page shot at an angle ends up looking like a flat scan.
Step 2. On-device recognition
A modern iPhone recognizes text locally, without sending the photo to a server — it's both faster and more private. The system finds text blocks in the image, splits them into lines and characters, and matches them against language models. The neural models of 2026 confidently read even handwriting and complex fonts.
Step 3. Understanding structure
Recognizing characters is half the job. Next you need structure: where's the heading, the paragraph, the table. For a translator this is critical: you translate sentences, not individual words, or the meaning falls apart.
Step 4. Translation and overlay
The recognized text goes to the translation model — our apps support 60+ languages. In Photo Translator's AR mode the translation is overlaid right on top of the original text on screen, preserving the size and position of the label.
The whole chain — from tapping the button to the translation on screen — takes less than a second.
Why this matters for privacy
Local processing means your documents and photos don't leave the device unless they have to. It's the principle we follow across all VOIO products: minimum data, maximum value.
