PDF to Audio — listen to any PDF, or save it as an MP3
Drop a PDF and listen to it read aloud by a natural-sounding voice, then save it as an MP3 for the car, the gym or the commute. The voice runs on your own device, so the PDF is never uploaded.
Fit Meter
Settings appear once you add a PDF.
Still going. Your own device is doing this, so a long scan or a big batch takes as long as it takes — and none of it is being uploaded.
Bytes of your files sent anywhere: 0 B · nothing leaves this tab
Every request this page makes is listed as it happens. You’ll see our own code loading, and for some formats a codec engine, but never your file going out.
- Waiting for network activity…
Check it yourself: open your browser’s developer tools on the Network tab, or turn off Wi-Fi once the page has loaded. The tools keep working.
How to convert a PDF to audio and save it as an MP3
-
Add your PDF
Drop your PDF. Its text is put in reading order, with the running heads, footers and page numbers that interrupt a printed page left out.
-
Check the settings
Pick one of five natural voices — American or British, a woman or a man — and a speed, then press play. Skip a paragraph or jump to any page as it reads.
-
Save the result
Press Save as MP3 for the whole PDF or just some pages. The audio is recorded on your device and saved like any other download.
Why it doesn’t upload
PDF-to-MP3 services upload your document and make the audio on their servers. Here the voice itself runs in your browser: the model is downloaded once, and from then on your text is turned into speech on your own device — which matters for the contracts, statements and study notes people most want to listen to on the move.
Check it in your browser’s Network tabWorth knowing
- The natural voices read English. A PDF in another language is read by your device’s own voice for that language, which can be played but not saved as an MP3 — browsers can’t record it.
- The first time, the voice is downloaded from Hugging Face, the open model library: about 93 MB, once. Your browser keeps it, so later visits start straight away and it works offline. Your PDF is not part of that download and is never sent.
- The voice is made by your device as it reads. On a phone or an older computer that can be slower than speech, so there may be short pauses between sentences, and saving a long MP3 can take longer than the audio lasts. Keep the tab open while it records; the screen is kept awake.
- A scanned PDF has no text to read — it is a picture of a page. That needs OCR, which is coming to the site but is not live yet.
- Tables and columns are read in reading order, which is not always the order a table is meant to be read in.
Under the hood
What happens to your file
The text is put into reading order without its headers and page numbers, then spoken by Kokoro, an open neural voice model running on ONNX Runtime in your browser — a few sentences at a time in parallel workers. For an MP3, the audio is encoded as it is made, so the whole recording is never held in memory at once.
Reference: Kokoro-82M model card
Questions
How do I convert a PDF to audio for free?
Drop the PDF here and press play to listen, or press Save as MP3 to keep it. There is no account, no watermark and no limit on length.
Can I download the audio as an MP3?
Yes. Save as MP3 records the whole PDF, or only the pages you choose, with the voice and speed you picked. It is a 64 kbps mono MP3 — clear speech at about 30 MB an hour — that plays on any phone, car stereo or podcast app.
Which voices can I choose from?
Five: Heart and Bella (American women), Michael (an American man), Emma (a British woman) and George (a British man). They come from Kokoro, an open neural voice model. The many robotic system voices a browser usually lists are left out on purpose.
Will it read the page numbers and headers too?
No. Text that repeats at the top or bottom of most pages — running heads, footers, page numbers — is recognized and skipped, and paragraphs split across a page break are read as one.
How long does saving an MP3 take?
It depends on your device, because your device makes the voice. A recent laptop records about as fast as the audio plays, sometimes faster; a phone is slower. The progress bar shows the time left after the first few sentences, and you can save a few pages at a time.
Is my PDF uploaded to be read aloud?
No. The text is taken out of the PDF in this browser tab and spoken by a voice model running on your device. The only download is the voice itself, the first time — nothing about your document goes with it.