How to Set OCR Languages, Auto-Detect and Correction on iPhone

TL;DR

In Settings pick the languages your documents contain (default is Simplified Chinese and English). Keep the list short; auto-detect still respects it. Turn language correction on for prose and off for names, codes and numbers.

The problem

OCR set up for English reads a Japanese contract as nonsense, and a bilingual report comes out with one language mangled. Turning on every language isn’t the answer either, because the recogniser has more wrong options to choose from.

Where the settings are

In Settings, the OCR section has:

  • Recognition Languages: open the list, search, tick the languages that appear in your documents.
  • Automatic Language Detection: guess the language per page. Off by default. Your selected languages are still applied.
  • Language Correction: on by default.
  • Recognition Level: Accurate or Fast.
  • Render Resolution: 200, 300 or 400 DPI.
DocumentsLanguagesNotes
English onlyEnglishCorrection on
Chinese + EnglishSimplified Chinese, English (default)Use 300–400 DPI for dense text
Mixed EuropeanThe 2–3 languages you actually seeAuto-detect on
Invoices, IDs, serialsMain languageCorrection off, so numbers aren’t “fixed”

Tips

  • Change settings, then re-run extraction on the project; results aren’t updated automatically.
  • Native-text pages ignore these settings, because they don’t need OCR.
  • If a page reads as gibberish, first check the language list, then the resolution.

FAQ

What is the default? Simplified Chinese plus US English.

Should I select every language? No. A focused list usually gives better accuracy. Selected languages still apply when automatic detection is on.

What does language correction do? It lets the recognizer fix likely mistakes using a language model. Good for sentences; it can alter names, codes and numbers.

How many languages are supported? Whatever the on-device recogniser supports on your iOS version. The picker shows the count.

The app PDFExtract →