PDF to speech

Importing the PDF

Drag the PDF onto the text box, or click Import text and choose the file. The text is extracted on your device and appears in the box, where you can read and edit it before anything is spoken. Nothing is uploaded, which matters for contracts, medical letters, coursework and other private documents.

If you only need part of the document, delete the rest from the box before generating. It is often quicker to import the whole file and trim it than to try to copy a few pages out of a PDF viewer.

Text-based versus scanned PDFs

A quick test: open the PDF in any viewer and try to select a sentence with your mouse. If you can highlight individual words, the PDF has real text and will import. If the whole page highlights as one block, or nothing highlights, it is a scanned image. Scanned PDFs have no text to read, so the box will come up empty or nearly empty.

For a scanned document, you need to run it through a text recognition step in another program first, then import the resulting file or paste the text. The tool does not do text recognition itself.

Cleaning up PDF text for listening

PDF text often carries leftovers from the page layout. Look for repeated headers and footers such as the document title or "Page 12 of 40" and delete them, since they will otherwise be read aloud every page. Footnote numbers stuck to the ends of words and hyphenated words split across lines are also worth a quick pass.

Two-column layouts, such as academic papers and newsletters, sometimes come out with lines interleaved from both columns. Skim the imported text; if paragraphs look jumbled, fix those sections before generating. Tables and figure captions rarely make sense as speech, so it is usually better to remove them or replace them with a short sentence summarizing the point.

Long reports and papers

There is no character limit, so a long report can go in as one piece. Playback starts while later parts are still being generated. For very long documents, consider splitting by chapter or section and downloading each as its own MP3, which makes it easier to pick up where you left off on a phone. See long text to speech for more on this.

For technical documents, try a speed of 0.9 so figures and terms have time to register. Numbers, percentages, dates and common units like km and MB are read in their spoken form automatically, which helps a lot with reports. Acronyms and product names may need respelling in the box if they come out wrong.

Listen now or save as audio

You can press play and listen in the page, click any part in the parts list to jump to it, or download the result as MP3 for a commute or walk. If you would rather follow along with text on screen, SRT subtitles are also available. For Word documents, the same approach applies; see Word to speech.

Frequently asked questions