PDF to audio
Convert a PDF to audio, free
Open a PDF and SpeakScribe reads its text aloud with an AI voice — 361 voices in 75 languages — and gives you the audio as MP3 or WAV. The PDF is opened by your own browser and never uploaded. You can try it without an account (up to 1,500 characters a try); a free account reads 5,000 characters at a time, about 5 minutes of English audio.
Choose a PDF to turn into audio
PDF with selectable text · up to 20 MB · read on your device, never uploaded
or drop the file here
- Pages
- Characters
- Listening time
Listening time at 16.0 characters a second, the median pace of the English voices we measured. Other languages differ — see the table below.
How it works
-
Step 1
Choose your PDF
Pick a PDF with selectable text (up to 20 MB). It is read in your browser — the file is never uploaded.
-
Step 2
Check the text
See the page count, characters and estimated listening time, then open the text in Voice Studio.
-
Step 3
Pick a voice and render
Choose one of the voices, set the speed and download the audio as MP3 or WAV.
How much of a PDF fits in one render
Each render reads up to a set number of characters, spaces included. Minutes are for English at the measured median pace; longer texts are rendered in parts and joined into one file.
| Plan | Characters per render | ≈ English audio |
|---|---|---|
| Free account | 5,000 | 5 min |
| Starter | 30,000 | 31 min |
| Pro | 70,000 | 73 min |
| Studio | 100,000 | 104 min |
| Agency | 150,000 | 156 min |
Reading pace by language
The same number of characters takes longer in some languages. Median pace of every voice in the language, each reading the same sample passage (measured October 2026).
| Language | Characters/s | 5,000 characters |
|---|---|---|
| English (75 voices) | 16.0 | ≈ 5.2 min |
| Hindi (6 voices) | 12.8 | ≈ 6.5 min |
| Bengali (4 voices) | 13.8 | ≈ 6.0 min |
| Urdu (4 voices) | 13.0 | ≈ 6.4 min |
| Persian (2 voices) | 13.2 | ≈ 6.3 min |
| Arabic (32 voices) | 9.5 | ≈ 8.7 min |
| Spanish (48 voices) | 15.7 | ≈ 5.3 min |
What the reader tidies up
- Lines that a PDF breaks in the middle of a sentence are joined back together, so the voice doesn't pause at every line end.
- Words split with a hyphen at the end of a line ("exam-" / "ple") are rejoined.
- Lines that contain only a page number are dropped.
- Paragraph breaks are kept, so the voice still pauses between paragraphs.
Running headers, footers and footnotes stay in the text. Delete them in Voice Studio before you render, or the voice will read them on every page.
PDFs that won't work
- Scanned PDFs — pages that are pictures of text have nothing to read. If you can't select a sentence in your PDF viewer, run OCR first.
- Password-protected PDFs — remove the password, or copy the text and paste it into Voice Studio.
- Very large files — up to 20 MB; only the first 500 pages are read.
- Tables and columns — text is read in the order it is stored in the PDF, which may not match two-column layouts. Check the text before rendering.
Choosing a voice for a long PDF
For an hour of listening, a calm and slightly slower voice is easier to follow than a bright, fast one. Every voice page shows the voice's measured pitch and pace, so you can compare before you render. For study notes and lecture handouts, see the guide text to speech for studying. Audio from Studio voices may be used for any purpose; Global voices are for your own content (Terms, section 6).