The paragraph you are hearing stays lit

Sandbook's reading view does not replace the page with a player. The text stays on screen while the voice moves through it, and the paragraph currently being read carries a soft tint, with the word being spoken marked inside it. The view scrolls on its own to keep that paragraph visible, so the place you are at is never something you have to go looking for. If you scroll away on purpose, a “Back to current” button takes you back.

That matters most in the moment attention slips. Look up, answer someone, look back: the marked word is exactly where the voice is, and you rejoin the text instead of restarting it. It is also how you catch the app mishearing a name or a piece of notation — you can see the words it is on.

Going back is a tap, not a scrub. Tap any paragraph and the voice starts from there, so re-hearing a difficult sentence costs one gesture rather than dragging a progress bar around and overshooting twice. You can also skip a paragraph at a time, move by chapter, or open the chapter or page list and jump anywhere in the file.

How the text is drawn is not a cosmetic preference for everyone, so the controls are in the reader's Reading Options rather than buried in Settings. There are four themes, Light, Sepia, Dark and Auto, so a page at night is not a white rectangle in a dark room. There are five fonts: System, Serif, Rounded, Mono and Legible, a Verdana-style face with open letter shapes. Text size runs from 0.8× to 1.6× on top of your system text size, with compact, normal or relaxed line spacing, and an immersive mode hides the controls.

Sandbook takes no position on which combination is better for you, because that would be a claim we cannot support. They are simply all there, they can be changed in the middle of a chapter, and the reader keeps whatever you last chose. The same is true of the voice: 41 voices, including 28 English voices (20 American and 8 British), every one of them free, and any of them can be auditioned before you commit an hour to it. Each book remembers its own voice and speed.

Speed runs from 0.5× to 2.5× in tenth steps, and the way it is implemented is the reason it is usable at the slow end. Rather than stretching a finished waveform, Sandbook scales the durations the model predicts for each sound, so 0.6× is a person speaking deliberately rather than a recording being dragged out. Nothing warbles, and nothing turns into a chipmunk at 2× either.

In practice that gives you two different tools from one slider: slow for a paragraph that is doing real work, and fast for the third pass through material you already know. Both are one tap from the reading view, and the setting survives closing the app.

Every file keeps its own reading position. Put a book down at chapter nine, open a PDF, come back a week later, and the book opens at the paragraph where you stopped, with no bookmarks to set, no syncing and no cloud library to log into. Your library shows how far through each file you are, and how long it takes to hear.

There is no account anywhere in the app, which also means there is no setup ritual between deciding to try this and hearing a page. You install it, open an EPUB or a PDF, and press play. It is free with no in-app purchase, so nothing has to be cancelled if it turns out not to suit you.

An app that lists features next to the word dyslexia should be explicit about where its claims stop, so here is the boundary in full.

• Sandbook makes no medical or clinical claim. It does not treat, diagnose, manage or improve any condition, and no study has been run on it.
• It holds no accessibility certification and is not audited against any accessibility standard.
• It is not a screen reader and not a replacement for VoiceOver, Speak Screen or any other iOS accessibility feature. It reads documents you open in it, and nothing else on your phone. The app itself has VoiceOver, Voice Control, Switch Control, larger text and contrast improvements.
• Scanned pages are recognized from the camera or photos, but an image-only PDF file is not, and handwriting is not promised.
• Voices cover six languages: US and UK English, Spanish, French, Italian, Brazilian Portuguese and Hindi.
• Real-time reading pauses when the screen locks, because the model holds real GPU memory while it works. A chapter prepared ahead of time can play with the screen locked.
• It needs an iPhone 12 or later on iOS 18.2, or an iPad with an A14 or M1 chip or later.

If any of what is on this page helps, it will help because of what the app does, not because of a label attached to it. That is the only claim being made here.