Skip to content
Limited-time offer. Save up to 50% on Speechly Premium. See pricing

Why Some PDFs Read Strangely (and How to Fix It)

Why some PDFs read strangely and how to fix it: what causes gibberish, jumbled columns and repeated page numbers, plus a 2-minute test and a fix for each.

Updated 9 min readBy the Speechly team
A woman looks puzzled at strips of a cut-up printed page scattered out of order on her desk, a phone beside them.
Quick answer

Some PDFs read strangely because a PDF stores where each letter sits on the page, not the order you read it in, so read-aloud tools have to guess. Scans, broken font data, columns, headers and line breaks make them guess wrong. Fix it with OCR, a different reading order, header and footer skipping, or a cleaner copy of the document.

Key takeaways
  • What you paste is what you hear: copy a paragraph into a plain text editor, and the paste shows exactly what the voice will read.

  • Silence usually means a scan with no text layer; gibberish means broken font data. OCR fixes both, because it reads the page image instead of the stored text.

  • Jumbled order comes from columns, sidebars and missing or wrong tags. Change the reader's reading order, or select one column at a time.

  • Repeated headers, page numbers and pauses at every line are layout leftovers. A reader that skips headers and footers, or a cleaned-up copy, removes them.

  • If you make PDFs, export them tagged from Word and set the document language, and they'll read well in almost any tool.

Here's why some PDFs read strangely and how to fix it, in short: a PDF records where each letter sits on the page, not the order you read it in, so every read-aloud tool has to guess. Scans, broken fonts, columns, headers and line breaks make it guess wrong. Each has its own fix, and most take a minute.

Below: what each symptom means, a two-minute test for any file, and fixes using free tools and Speechly's PDF reader.

Quick diagnosis: what you hear and what's causing it

Match what you hear to its cause, then jump to the fix.

What you hear Most likely cause Fix
Nothing, or "empty page" Scanned PDF with no text layer Run OCR
Gibberish, symbols, wrong letters Fonts with no character mapping OCR the page images, or get the original
Sentences jumping between columns or boxes No tags, wrong tags or a complex layout Change the reading order, or select one column
The title and page number on every page Headers and footers not marked as page furniture Skip headers and footers, or delete them
A pause at every line, words split in two Line breaks and end-of-line hyphens Use a reader that joins lines, or reflow the text
Words run together or spelled out letter by letter Missing or extra spaces in the text layer OCR the file, or ignore it if it's only headings
An English voice mangling another language Missing or wrong document language Pick a voice in the document's language
Nothing, though you can select the text Security settings that block accessibility Ask the sender for an unrestricted copy
Tables, equations and footnotes read as noise Content with no natural spoken form Skip them, or listen to an HTML version

Why read-aloud tools have to guess with PDFs

A PDF is built to look identical everywhere, like a printout. Inside, it's mostly drawing instructions: these letter shapes, in this font, at these coordinates.

What it usually leaves out is meaning. Microsoft's Word team puts it plainly: most PDFs don't include information about structural content elements such as paragraphs, tables or columns.

So a read-aloud tool rebuilds the text from positions: which letters form a word, which lines form a paragraph, which block comes next. On a simple page it gets this right. On a journal article or a brochure, it often doesn't.

Tagged PDFs are the exception. Tags are a hidden outline: "this is a heading, this paragraph comes next, this is a page number." The W3C notes that reading order is determined primarily by the tag order, and that when a two-column page isn't tagged properly, a screen reader may read straight across both columns. Tags help only when they're right.

How to tell what's wrong with your PDF in two minutes

Read-aloud tools speak the same text you get when you copy, so a paste shows you what the voice will say.

  1. Try to select a word on the page. If nothing highlights, it's a scan: go to the OCR fix below.
  2. Select a full paragraph, plus the line above and below it, and copy it. If the text highlights but won't copy, the file may be locked (cause 8).
  3. Paste it into a plain text editor: TextEdit on a Mac (choose Format > Make Plain Text first), Notepad on Windows, or a note on your phone.
  4. Read the paste. Garbled characters, glued-together words, a stray page number or a line from the next column all show up here, before you've spent 20 minutes listening.

If the paste looks clean but the voice still sounds wrong, the problem is the reader or its settings, not the file.

Why some PDFs read strangely and how to fix it, cause by cause

1. Silence: the PDF is a scan

A scanned PDF is a stack of page photos. The W3C says assistive technologies cannot read or extract the words in a scanned document, so the reader finds nothing to say.

Fix: run OCR (optical character recognition), which turns the page images into real text. Free options include Chrome on a computer, Google Drive and Adobe's online tool; the steps are in how to listen to a scanned PDF. OCR can misread words, so spot-check a page against the original.

2. Gibberish: the fonts can't be decoded

The page looks perfect, which is what makes this one confusing. The PDF draws its letter shapes correctly but has no table mapping those shapes back to real characters, so the stored text is junk. A better voice can't help, because it only gets the junk.

Fix: ignore the stored text and read the picture instead.

  • One page at a time: take a screenshot and use an app that reads text from an image aloud.
  • The whole file: the free, open-source OCRmyPDF tool's --force-ocr option rasterizes the pages and runs OCR on the images, replacing the broken text. Without it, OCRmyPDF stops with a "page already has text" error, and other OCR tools may also keep the gibberish.
  • Best: if someone sent you the PDF, ask for the original Word file or Google Doc.

3. Jumbled order: columns, sidebars and text boxes

Guessing from positions, a reader can go from line one of the left column to line one of the right. Captions, sidebars and footnotes get slotted in mid-sentence, and badly tagged files jump around too.

Fix: change how your reader works out the order, or feed it one column at a time.

  • Adobe Acrobat Reader: go to Edit > Preferences > Reading (Windows) and try the Reading Order options. Infer Reading Order From Document works out columns from the layout; Left-To-Right, Top-To-Bottom Reading Order goes strictly by position. Adobe recommends ticking Override The Reading Order In Tagged Documents only for poorly tagged files.
  • Mac, iPhone and Android: select a single column and play just that selection with the built-in speak feature.
  • Research papers: listen to a web version. arXiv's HTML link appears on abstract pages below the PDF download link, though a small share of papers have none.

Full steps are in how to listen to two-column PDFs.

4. Repetition: headers, footers and page numbers

You hear "Annual Report, page 14" mid-sentence because nothing in the PDF says that line is page furniture. Well-made PDFs mark running headers and footers as pagination artifacts so readers skip them. The W3C technique for headers and footers notes that Word does this when you use its header and footer areas, but many PDFs are made without it.

Fix: use a reader that skips repeating headers and footers, or convert the PDF to a document and delete them. How to clean up a document before listening to it shows the quickest way.

5. Choppy reading: line breaks, hyphens and ligatures

Every printed line is usually stored as a separate piece of text. Some readers treat each line end like a full stop, so you hear a pause every eight or ten words.

Two more leftovers split words:

  • End-of-line hyphens: "infor-" on one line and "mation" on the next are read as two fragments.
  • Ligatures: many fonts draw "fi", "fl" and "ff" as one joined shape. If that shape isn't mapped to its letters, "find" can come out as "nd".

Fix: use a reader that joins lines into sentences before speaking, or reflow the text. Opening a PDF in Word converts it into an editable document with real paragraphs; Microsoft says this works best with files that are mostly text. Then use Find and Replace to remove leftover line-end hyphens.

6. Spacing: words glued together or spelled out

Spaces in a PDF are often just gaps, so the reader decides where words end. Tight justified text can lose its spaces ("thequickbrownfox"), and letter-spaced headings can gain them ("C H A P T E R") and get spelled out.

Fix: if the paste test shows this on most pages, OCR the file as in cause 2; OCR rebuilds the spacing from the image. If it's only headings, ignore it.

7. Wrong pronunciation: the language is missing or wrong

A PDF can declare its language. The W3C says that when it does, screen readers can load the correct pronunciation rules. Without it, a French report might be read with English rules.

Fix: pick a voice in the document's language yourself. If you own the file and have Acrobat Pro, set the language under File > Properties > Advanced > Reading Options > Language.

8. Nothing reads, but the text selects: the file is locked

PDF security settings can stop assistive technology from reaching the text. In Acrobat or Acrobat Reader, open File > Properties > Security and look for Content Copying for Accessibility in the restrictions summary. If it says Not Allowed, some readers will stay silent.

Fix: ask whoever sent the file for a copy without that restriction. Content copying and accessibility copying are separate permissions, so they can keep copying blocked and still let read-aloud tools in.

Also: tables, math and footnotes

Even a perfect PDF has content with no natural spoken form: tables read cell by cell, equations as symbols, footnote numbers glued to words. See how text to speech handles math, tables and footnotes.

When to skip the PDF and use another version

Often the best fix is a better source:

  • The original document: a Word file or Google Doc keeps its paragraphs, headings and language.
  • A web version: journal articles, government reports and arXiv papers often have an HTML page that reads top to bottom.
  • An EPUB: eBooks reflow, so they have no fixed pages, columns or running headers.
  • An accessible version: ask the sender for a tagged PDF or another format.

How to listen to a messy PDF with Speechly

Speechly is a text-to-speech reader for iPhone, iPad, Android, Mac and Windows that reads PDFs, Word documents, eBooks and web pages aloud with 25+ natural AI voices. It can't repair a broken text layer, but it makes problems easy to spot and skip.

  1. Open the PDF in Speechly on iPhone, iPad, Android or Mac.
  2. Press play and watch the word-by-word highlighting on the first page; it shows exactly where the text jumps or breaks.
  3. If it goes wrong, jump to the sentence where you want to continue, and reading starts there.
  4. On iPhone and iPad, Speechly can skip repeating headers and footers, so the title and page number aren't read on every page.
  5. Set the speed anywhere from 0.5× to 2×. Your place is saved and picks up on any device you sign in to.

Have the paper copy? Take a photo of the page in Speechly and listen to it. That skips the broken PDF entirely.

On a computer: select a clean passage in any app and press ⌘E on a Mac or Alt+R on Windows to hear it in a floating player.

The free plan includes two files with premium voices, no card required. Premium is $14.99/month or $139/year for unlimited listening on every device.

If you make PDFs: how to stop them reading strangely

Most reading problems are set when the PDF is created. Four habits fix them for every listener:

  1. Use heading styles and the header and footer areas, not text boxes.
  2. Export a tagged PDF. In Word for Windows, choose File > Save As, pick PDF, select Options and make sure Document structure tags for accessibility is ticked. In Word for Mac, choose Best for electronic distribution and accessibility, which uses a Microsoft online service.
  3. Set the document language before you export.
  4. Export, don't scan, and leave accessibility copying allowed if you add security.

Then run the paste test on your own file. If the paste reads cleanly, the voice will too.

Frequently asked questions

Why does my PDF read aloud as gibberish?

The PDF's fonts don't map their letter shapes back to real characters, so the page looks fine but the stored text is junk. Run OCR on the page images, which ignores the broken text, or ask for the original document.

Why won't my PDF read aloud at all?

Usually it's a scanned PDF: pictures of pages with no text layer. If you can't select a single word, run OCR first. If you can select text but nothing reads, check whether the file's security settings block accessibility.

Why does Read Out Loud pause at the end of every line?

Each printed line is stored as a separate piece of text, and some readers treat every line end like the end of a sentence. A reader that joins lines into sentences, or a reflowed copy of the text, reads smoothly.

Why does text to speech read the page number and header on every page?

The PDF doesn't mark them as page furniture, so the reader treats them as part of the text. Use a reader that skips repeating headers and footers, or delete them from a converted copy.

Can a better voice fix a PDF that reads strangely?

No. The voice reads whatever text the PDF hands it. A natural voice makes good text pleasant to hear, but broken order or garbled characters have to be fixed in the text first.

How do I make a PDF that reads aloud correctly?

Write it in Word or another editor with real headings, put running text in the header and footer areas, set the language, and export a tagged PDF. In Word for Windows, tick Document structure tags for accessibility under Options when you save as PDF.

Speechly

Listen with Speechly

Natural AI voices on iPhone, iPad, Android, Mac and Windows, with word highlighting and your place saved on every device. Free to download.

Prices and features of other products were checked on November 18, 2026 and can change. Speechly is not affiliated with the other companies named here.