Built for readers who find reading itself the hardest part

A personal starting point

I'm dyslexic. Sustained silent reading — the kind a 600-page technical textbook demands — is exhausting and genuinely painful for me in a way it isn't for most people. Words swim. Lines get lost. A single dense paragraph about, say, SDRAM timing parameters can take several passes to actually land.

Listening is different. Spoken language is something I process well, fluently, for hours. It's how I actually get through dense material — podcasts, audiobooks, lectures. The problem is that technical books were never written to be read aloud. They lean on diagrams, register tables, and equations that a plain text-to-speech pass either skips entirely or reads out as meaningless noise ("t sub C K equals one over f...").

So the book stayed unread, or I read it slowly and painfully, or I found some workaround that lost half of what mattered. Firebrat exists because none of those were good enough.

Who this is for

Dyslexic readers, first and specifically. But the same design choices help a much wider set of people:

• People with low vision, who benefit from audio as the primary channel and adjustable, high-contrast text as the secondary one.
• People with ADHD, for whom short narrated segments with a visible, tappable structure are easier to stay engaged with than a wall of unbroken text.
• Anyone who processes spoken language more fluently than dense printed technical writing — which, it turns out, is a lot of people, whether or not they'd describe themselves as having a reading difficulty.
• Anyone who just wants to get through a technical book on a commute, a walk, or while doing something else with their hands, without losing the diagrams that make the book worth reading in the first place.

What "accessible by design" actually meant here

Concretely, in the app itself, not as an afterthought:

• Adjustable text size is a first-class setting, reachable from the reader screen itself, not buried in a system-wide accessibility menu.
• Segments are short — one sentence at a time — so the highlighted text on screen never becomes another wall to parse; it's always a small, current, matched-to-what-you're-hearing chunk.
• High-contrast highlighting for the segment currently being spoken and the formula currently being described, not a subtle color shift that's easy to miss.
• Manual navigation always works, regardless of autoplay state — nothing about the reading flow is a one-way street you can get stuck in or have to "catch up" with.
• Figures never disappear. The one being talked about is emphasized, but every other diagram for the section stays visible and tappable — because sometimes you want to compare two figures, or you already know which one you need, and a rigid one-at-a-time flow would fight you.

An invitation, not a lecture

I'm not trying to speak for every dyslexic reader, or every person with a different relationship to text than mine — disability and reading difficulty aren't monolithic, and what helps me won't help everyone identically. What I can say is that this is a real tool, built to solve a real, specific, personal problem, released under a permissive-enough copyleft license so that anyone who wants to point it at their own technical book — or fork it and adjust it for how they read best — can.