Mind's Ear: a free e-book reader that reads aloud, in your browser

Mind's Ear

An e-book reader that reads to you, right in your browser.

An e-book reader that reads to you, right in your browser

Mind's Ear opens your EPUB, PDF and text files as a book, with pages to turn, a table of contents and your choice of font and theme. Click any sentence and it reads aloud from there in a natural-sounding voice, highlighting each word and turning the pages as it goes. The AI voice model runs on your own device, so it's free, private and has no limits.

How it works

  • Click a sentence to hear the book from there, and Click it again to pause.
  • Turn the page with the arrow keys, a swipe, or a click at either edge. On a touch screen, swipe down on the page to go back to the library.
  • Turn the page yourself while it reads, and Jump back takes you back to the voice.
  • Drag across a passage, or press and hold a word, to highlight it in one of five colours or copy it. Your highlights are listed in Contents.
  • When the controls have faded, move the pointer to where they were, or click off the text, to bring them back.
  • Drag along the line above the play button to go anywhere in the book.
  • The voice is to the left of the play button, and the speed to its right.
  • Reading settings, at the top right, has the font and its size, the page colour and the margins.
  • Pause it, or skip a sentence back or on, with your keyboard's media keys, your headphones, or your phone's Lock Screen, where your browser offers them.
  • Add books with Add a book, or drop them anywhere on the page.

Features

  • Your books, as books

    Open EPUB, PDF and text files as pages to turn, with a table of contents, two pages side by side on a wide screen, and your choice of font (including OpenDyslexic), text size, page colour and margins.

  • Click a sentence to listen

    Every sentence highlights under your pointer. Click one and it's read aloud from there, highlighting each word and turning the pages as it goes.

  • A library that remembers

    Books you open stay on your shelf, with their covers and how far you've read. Open one again and you're back on the same page.

  • Highlights

    Press and hold a word (or drag across a passage) to highlight it in one of five colours, or copy it. Your highlights are listed in Contents, to go back to.

  • Four classics to start

    Alice's Adventures in Wonderland, The Adventures of Sherlock Holmes, Frankenstein and The Wonderful Wizard of Oz are on your shelf from the start, from Project Gutenberg.

  • Private by design

    Your books never leave your computer. They're read and kept in your browser; nothing is uploaded or logged.

  • Natural AI voices

    54 voices from the Kokoro model, including American and British English, plus accented voices from seven other languages.

  • Completely free

    No account, no subscription, no limits. The voice model runs on your own device, using your GPU through WebGPU when available.

Frequently asked questions

›Which files can I open?

EPUB e-books, PDFs and plain text (.txt) files. EPUBs keep their chapters, contents, italics and pictures; PDFs and text files are reflowed into pages, with their chapters found where possible. Scanned PDFs have no text to read, so they would need OCR first.

›Can I open Kindle books?

Only DRM-free ones. Books bought from the Kindle store are DRM-protected and can't be opened by any other reader. Many DRM-free books (such as those from Project Gutenberg or Standard Ebooks) are available as EPUB.

›Are my books uploaded anywhere?

No. Your books are read and stored in your browser, and the text-to-speech model runs on your device, so they never leave it. The only downloads are the speech model and voices (from Hugging Face) and the engine that runs them (from jsDelivr), which are the same for everyone. There are no analytics or trackers.

›Where are my books kept?

In this browser's storage on this device, along with how far you've read. They aren't synced to other devices or browsers, and clearing the site's data removes them.

›Why does the first read take longer?

The first time a book is read aloud, your browser downloads the AI voice model: about 325 MB, or 92 MB on devices with little memory and no WebGPU. It's cached afterwards, so later visits start in a moment.

›Which browsers does it work in?

Any modern browser. It's fastest in browsers with WebGPU, such as Chrome, Edge and Safari 26 (including on iPhone and iPad), where it uses your graphics chip. Other browsers run the model on the CPU, which is slower.

›What languages and voices are available?

It reads English text, in 28 American and British English voices plus 26 voices from Spanish, French, Hindi, Italian, Japanese, Portuguese and Mandarin speakers who read English with their native accent.

Credits

Mind's Ear is made possible by these open-source projects. Thank you to their authors.

  • Kokoro-82M

    The text-to-speech model and its voices, built on the StyleTTS 2 architecture. By hexgrad · Apache 2.0

  • Kokoro-82M ONNX

    The model converted to run in browsers, with word timings for highlighting. By ONNX Community · Apache 2.0

  • kokoro-js

    Runs Kokoro in JavaScript. By hexgrad and Xenova · Apache 2.0

  • Transformers.js

    Loads the model and runs it in the browser. By Hugging Face · Apache 2.0

  • ONNX Runtime Web

    Runs the model on your GPU (WebGPU) or CPU (WebAssembly). By Microsoft · MIT

  • phonemizer.js

    Turns text into speech sounds, using eSpeak NG. By Xenova · Apache 2.0

What your browser downloads

The first time a book is read aloud, your browser fetches the following directly from these sources and caches them. Your books are never part of these requests.

  • Speech model (about 325 MB, or 92 MB on devices with little memory; once)Hugging Face
  • Each voice (about 0.5 MB, the first time it's used)Hugging Face
  • ONNX Runtime engine (about 20 MB, once)jsDelivr