Skip to content

Translate selected text or a page

New in v0.10.0

On-device translation ships in v0.10.0.

When to use this

You want a quick reading aid for selected PDF text or the page you are on, without changing the document.

  1. On macOS 15 or later, optionally select text on the PDF page.
  2. Choose View → Translate Selection. If no text is selected, choose View → Translate Current Page instead. The same action also appears under More (⋯) when that toolbar button is visible.
  3. Pick the target language. Orifold shows the original and translation side by side.
  4. Use Copy if you need the result elsewhere, then choose Done.

Translation is deliberately read-only. It never replaces text, adds an annotation, or writes the result back to your PDF.

Before the first translation, Orifold explains the boundary and asks you to continue. It uses Apple’s Translation framework. macOS may need to download the selected language model from Apple once; after the required model is installed, the document text is translated on this Mac.

Note

Translation requires macOS 15 or later, even though the rest of Orifold supports macOS 14. The menu item is not shown on macOS 14.

  • A non-empty PDF text selection takes priority.
  • Without a selection, Orifold reads the current page in detected reading order.
  • Scanned pages need an OCR text layer first. If the page has no readable text, the translation action is unavailable.
  • Current-page translation reflects committed in-place text edits, not stale text from the original PDF.

Long passages are split at sentence and word boundaries, translated in order, and reassembled in the result panel.

Your document is not uploaded to an Orifold server—there is no Orifold server. The translation itself uses Apple’s on-device language models. The only network activity associated with this feature is the system-managed language-model download that macOS may request before a language is ready.

OCR scanned pages

Add a local text layer before translating a scan.

Read aloud

Listen to document text with macOS voices.

Privacy & local-first design

The exact local-processing and network boundaries.