Speech to text for dyslexia

How dictation gets around the spelling wall

Here's the mechanic, plainly. When you type, you are spelling every word yourself, in real time, while also trying to hold a thought together. For a dyslexic writer that's two demanding jobs stacked on top of each other, and the spelling job keeps interrupting the thinking job. Dictation removes the first job. You say "necessary," and the screen shows "necessary," spelled right, without you having to decide whether it's one C or two.

This is the part I want to be honest about, so I'll say it once and plainly: I'm a software architect, not a clinician, and nothing here is medical advice. Speech to text is an assistive writing tool. It does not treat or cure dyslexia, and anyone selling it as "clinically proven" to do so is selling you something. What it does is let written output better match spoken ability, which many people with dyslexia find genuinely freeing. If you want the research and the careful version, the Yale Center for Dyslexia & Creativity (dyslexia.yale.edu) and Understood (understood.org) both cover assistive tech well.

So the framing for the rest of this page is simple. We're not fixing anything. We're getting the spelling out of the way so the writing can happen.

Writing that matches how you talk

There's a second thing dictation does, quieter than the spelling fix but just as useful. It lets you write in your own voice. Plenty of dyslexic writers learn to shrink their vocabulary on the page, sticking to words they're confident they can spell, because a misspelled word feels worse than a smaller one. The richer word stays in your head where the spell-checker can't embarrass it.

When you dictate, that self-censoring stops mattering. You can say the longer, sharper, more exact word, because you only have to pronounce it, not spell it. The text that comes out reads more like you sound, which for a lot of people is the first time their writing has. That's not a small thing. It's most of the reason I think this tool matters for this audience specifically.

The fastest way is a system-wide hotkey

The smoothest setup, once you're past the free built-ins, is a desktop app that works like a keyboard. Whisper by Remskill isn't a browser extension or a single-app tool. It sits over everything: press a hotkey, speak, and the transcript is pasted wherever your cursor is. An essay in your word processor, an email, a note, a homework reply, a form. Same key everywhere, so there's nothing new to learn per app.

Getting set up is short:

That's the loop. The text appears in the document. You read it, fix what you want, keep going.

Speak, and it's spelled for you

This is the moment that does the work. You hold the key, you say a sentence the way you'd say it out loud to a friend, and a second or so later it's on the page, spelled correctly. Not your best guess at the spelling. The actual spelling. The thing that, with a keyboard, would have cost you three tries and a red squiggle, just appears.

Because the transcription runs on your own machine (pure-Rust, no Python sidecar, no server in the loop), it works offline, and the words never leave your laptop. For a student writing a personal essay, or anyone dictating something they'd rather not send to a stranger's server, that privacy is the default, not an add-on.

AI cleanup for punctuation and tidy-up

Spoken language is messy, and dyslexia has nothing to do with that part: everyone says "um," restarts sentences, and forgets where the commas go when they talk. Whisper has an optional AI step that tidies the punctuation and trims the filler before it pastes. So "um, so the, the trip is on Friday, I think, yeah Friday" becomes "The trip is on Friday." You speak loosely; the text reads cleanly.

That cleanup runs locally through Ollama, free, on your own machine. It's a genuine help here, because punctuation and sentence boundaries are exactly the mechanics that can trip a dyslexic writer up after the words are down. You can also turn it off and paste the raw transcript. Both are fine. Use whichever leaves you doing less hand-correction.

It stays on your computer

I'll keep this short because it's simple. In local mode, the audio and the transcript stay on your machine. Nothing is uploaded, nothing is logged on a server, nothing needs an internet connection. For a kid's homework, a private journal entry, or an adult drafting something personal, that's the right default. Your words are yours.

Windows' own built-in voice typing needs an internet connection to work at all. Whisper's local mode does not. If you're somewhere with patchy wifi, or you just don't want your writing leaving the room, local mode keeps working.

What the full Whisper app looks like

The hotkey is the part you'll use every day, but there's a settings surface behind it. You pick your transcription engine: Whisper models, whose multilingual variants cover 99 languages and whose English-only builds cover exactly one, or NVIDIA's Parakeet, about 600 MB and 5 to 10 times faster than Whisper on CPU, covering English plus 24 European languages. You set your hotkey, manage history, and save presets. None of that is required to dictate one sentence. It's there for when you want to tune.

Free built-ins worth trying first

Before you install anything, try what's already on your computer, because for a lot of people it's enough. On Windows, press the Windows logo key + H and voice typing opens in any text box, no subscription. On a Mac, the built-in Dictation shortcut does the same, and on Apple silicon it runs on-device. On a Chromebook or in Google Docs, Voice Typing under the Tools menu is free and surprisingly good. On a phone, the keyboard's microphone icon dictates into almost anything.

If you're a parent setting this up for a child, start here. Google Voice Typing in a Docs file is free, needs nothing installed, and is a low-stakes way to find out whether dictation clicks for them at all. The one tradeoff worth knowing: Windows' Win+H and Google Voice Typing both need an internet connection, while Mac dictation and Whisper's local mode don't.

Where a dedicated app pulls ahead is the rest of it: working in every app instead of one, the punctuation cleanup, dictating offline, and not having to re-open a voice-typing panel every time. The longer the writing sessions, the more those add up. For a single short note, the free built-in is the right call, and I'd rather you knew that than installed something you didn't need.

What Whisper costs

The local dictation pipeline, transcription and the AI cleanup over Ollama, is free for any signed-in user, with no card at signup. So getting your voice onto the page with Whisper costs nothing. Whisper Pro adds the cloud features (OpenAI transcription, cloud AI enhancement, voice web search) on a separate trial. The exact numbers live on the pricing page rather than here, because prices move and a blog post is a bad place to keep them current.

Spelling was never the same thing as having something to say. For a lot of people those two got tangled up early, in a classroom, with a red pen, and stayed tangled. Speech to text pulls them apart again: the saying stays yours, the spelling becomes the computer's problem. It won't undo any of the hard history. It just means the next thing you write can sound like you, the first time, without the wall.