How to read a PDF aloud on a Mac

Built in: Preview plus Speak Selection

1. Open the PDF in Preview. Double-click the PDF; it opens in Preview by default.
2. Select the text. Drag to select a passage, or click in the text and press Command-A to select everything.
3. Start speaking. Press Option-Esc (with Speak Selection on), or choose Edit → Speech → Start Speaking.
4. Stop. Press the shortcut again, or choose Edit → Speech → Stop Speaking.

If you haven't turned on Speak Selection yet, follow this guide first. Edit → Speech works without it.

Common problems

It reads headers, footers and page numbers

Speak Selection reads exactly what's selected, so in a long report you'll hear "Annual Report, page 14" between paragraphs. With the built-in tools the only fix is selecting smaller chunks.

It's a scanned PDF and nothing can be selected

A scan is a picture of text. On recent macOS, Live Text may let you select text in Preview; if it does, you can then use Option-Esc. If it doesn't, you need text recognition (OCR) first.

Hyphenated words and broken lines

Words split across lines ("informa- tion") can be read as two words, and paragraphs that cross a page break get a pause in the middle.

Two-column layouts

Reading order follows how the PDF stores its text, which isn't always the order you'd read it. This is a limitation of the file, and affects most readers, including Aloud.

Reading PDFs with Aloud

1. Drag a PDF into Aloud's reading window, choose File → Open PDF or Image…, or in Finder right-click the file and choose Services → Read File with Aloud.
2. Aloud opens it in a reading window and starts reading, highlighting each word. Running headers, footers and page numbers are left out, hyphenated line breaks are rejoined, and paragraphs that cross a page read as one.
3. Scanned pages are recognised on your Mac, so they read like any other page.
4. Use Go to Page to jump, click any sentence to start there, and reopen it later from the menu bar to continue from the sentence where you stopped.
5. Citation markers like [12] can be skipped, and you can save the whole reading as an .m4a file.