A Speak Screen alternative for whole books and PDFs

Speak Screen is not an app. It is an accessibility setting — in iOS 26 it lives under Settings, Accessibility, Read & Speak — that makes the whole operating system speak. Swipe down with two fingers from the top of the screen and iPhone reads what is on it, in any app that puts text there: your mail, a message thread, a page on a site you are signed in to, a PDF a viewer is displaying. It is free, it asks for nothing, and synthesis runs on the device once a voice is downloaded.

Sandbook is a reader. You bring it an EPUB, a PDF, a web page, a scanned printed page or some pasted text, and it keeps those as a library: a chapter picker, word-by-word highlighting, tap-a-paragraph-to-read, auto-advance between chapters, and each book opening where you stopped. It speaks with Kokoro-82M, an open-weight neural model bundled in the download and run on the GPU through Apple MLX at 24 kHz.

These are different jobs. Most people who install Sandbook should keep Speak Screen switched on, and this page will say so again at the end.

Said first, because it is free, it is already on your phone, and you should try it before installing anything:

• It works everywhere. Any app that draws text on screen can be read aloud, without that app having to support anything. No reading app can match that reach, and Sandbook does not try to.
• It costs nothing and counts nothing. No install, no account, no purchase, no quota.
• It runs on the device. Choose Default or Enhanced Quality; an Enhanced voice downloads once over Wi-Fi — Apple notes these can be 100 MB or larger — and after that synthesis is local.
• It highlights as it reads. The highlight style, colour and granularity are configurable for in-app text that is spoken aloud.
• Speak on Touch and a pronunciation dictionary. Tap the passage you want read, adjust the rate, skip forward and back, and dictate or spell out how a word should be said.
• Personal Voice. iOS can build a synthesized voice from your own recordings, processed on the device, and use it with Read & Speak. Sandbook has nothing comparable — there is no voice cloning in it at all.

Speak Screen reads what an app has already put on screen. That one sentence explains everything it does not do, and none of it is a flaw — it is the design. It has no library, no document importer, no EPUB or PDF pagination, no reading position kept per file, and no OCR of its own. If an app cannot display the text, Speak Screen cannot read it.

Over an article, that costs you nothing. Over a 400-page book, it is the whole experience: you keep the book open, scroll it yourself, and find your place again tomorrow. What a reader adds is the bookkeeping.

• A library you import into. DRM-free EPUB, PDF, TXT, RTF, Markdown and HTML from the Files app or the share sheet, web pages from Safari, and printed pages scanned with the camera. The books stay in the app.
• A position per book. Close it mid-chapter, open it next week, carry on. There is also a chapter and page picker and auto-advance to the next chapter.
• Read-along inside the document. Each word is marked as it is spoken, with auto-scroll, and any paragraph can be tapped to read from there, which is useful for reading along with the voice rather than instead of it.
• PDFs repaired before they are spoken. Sandbook rejoins the invisible line breaks a page layout leaves in the text and stitches hyphenated words back together, so a two-column paper reads as prose rather than line by line.
• A different kind of voice. 41 voices in six languages, 28 of them English, from a neural model bundled in the app, at 24 kHz, with speed from 0.5× to 2.5× applied inside the model so pitch is preserved rather than resampled.
• M4A export from the Reader tab. Whole books cannot be exported.

• You want text read where it already is — in mail, in messages, in an app Sandbook has no way to import from.
• You read in a language Sandbook has no voice for. Sandbook covers US and UK English, Spanish, French, Italian, Brazilian Portuguese and Hindi. Open Read & Speak on your own phone and look at the voice list there: whatever iOS has for your language is what Speak Screen will use.
• You listen with the screen off or the phone in a pocket. Sandbook's real-time reading pauses when the screen locks, because the speech model holds real GPU memory while it runs; only a chapter prepared ahead of time plays locked. If this is what decides it for you, test it in both before you choose.
• You use Personal Voice, or you need a pronunciation dictionary. Sandbook has neither.
• Your iPhone is older than an iPhone 12. Speak Screen asks for nothing extra; Sandbook's voices need an iPhone 12 or later because the model runs locally on that generation's GPU.
• You need a scanned, image-only PDF file read aloud. Speak Screen has no text recognition of its own, and Sandbook recognizes camera scans and photos but not PDF files.

If the voice is what you are weighing, you can try Sandbook's voices online before installing anything, and play the same sentence through Speak Selection to compare.