Read any text aloud, free

1. Paste or type your text in the box. A sample is already there, so you can simply press Start to hear how it sounds.
2. Pick a language and a voice. The round play button next to the voice plays a recorded sample at once, with nothing to load.
3. Press Start. The first time, your browser downloads the voice engine once and keeps it; then the reading begins by itself. After that the button says Read aloud and answers in seconds.

The voice is Kokoro, an open neural speech model, and it runs inside this browser tab rather than on a server. Your text is split into sentences and each one is turned into speech in turn, so the first sentence plays while the rest are still being made. You can slow the voice down to 0.5× or speed it up to 2.0×; the pace is changed inside the model, so a fast voice still sounds like a person talking quickly.

An email you would rather not read on a screen, an article, a page of notes, a speech to rehearse: all of it is read the same way.

• Short text, up to 1,500 characters: spoken in one go, with a player you can pause and scrub, and a Download WAV button.
• Long text, up to 100,000 characters: shown as paragraphs and read from the first one. The paragraph being read is highlighted, and tapping another paragraph jumps there.
• Files: Import a file opens an EPUB, PDF, Word (.docx), Markdown, HTML or plain-text file in a full-screen reading view and reads it from any sentence. There are separate pages for opening an EPUB online and for reading a PDF aloud.
• Voices: 41 in all, in US and UK English, Spanish, French, Italian, Brazilian Portuguese and Hindi. You can hear every voice before you choose.

If the text is your own writing, the essay reader reads it back one sentence at a time, so you can fix what you hear. If you need an audio file of a short text rather than a listen, the Kokoro TTS online page is the same engine set up as a voice generator.

There is no account, no trial and no counter. It can be free because nothing runs on a server: your own device does the work. That is also where the limits come from.

• The first visit downloads the voice engine once: 326 MB on a computer whose browser has WebGPU, 92 MB otherwise and on phones. Later visits load it from the browser in seconds.
• It is slower on phones. A phone makes speech more slowly than it plays, so expect a wait before the first sentence and pauses between sentences. Keep the screen on and the tab in front: reading stops when the phone locks.
• On a computer without WebGPU, the voice is also made more slowly than it plays.
• Long text and files are read aloud on the page and have no audio download. Short text downloads as a WAV file.
• It reads text you paste or a file you open. It does not open a link for you, so copy the article's text and paste it.
• Your text is not kept once you close the tab. For a file you open, only the place where you stopped is remembered, in this browser.

Privacy: your text and audio never leave your browser. A file you open is read on your device and is never uploaded.

This page is for the text in front of you right now. For a whole book there is Sandbook, the app: it reads EPUB books, PDFs, Word documents, web pages and scanned pages aloud on iPhone and iPad with the same 41 voices, marks each word as it is spoken, and opens every book at the paragraph where you stopped. Everything stays on your device, and it works offline. If you only want the phone to read what is on its screen, iPhone's built-in text to speech can do that.

The app is free, with no subscription and no in-app purchase. It needs an iPhone 12 or later, or an iPad with an A14 or M1 chip or later, on iOS 18.2 or later. One limit to know before you install: real-time reading pauses when the screen locks, so to listen with the screen locked you prepare the chapter's audio first.