What dictation actually is
If you've never used it, voice dictation is simply this: you talk, and your words appear as written text where you'd otherwise be typing. That's the whole idea. There's no special vocabulary you have to learn and no machine you have to talk to in a stilted way. You speak the way you'd explain something to a person, and the text shows up in your email, your notes, your messages — wherever your cursor happens to be.
What makes it worth learning is that talking is the fastest, most natural way humans produce language. You've been doing it your entire life, and it costs you almost no effort. Typing, by contrast, is a learned physical skill that's always a little slower than your thinking — much slower on a phone. Dictation lets you skip that bottleneck. This beginner's guide to voice dictation is meant to get you from never having tried it to comfortably using it, without the usual false starts.
How it works, briefly
Under the hood, a speech recognition system listens to the sound of your voice and converts it into words. Modern systems are good at this because they've effectively "heard" enormous amounts of human speech and learned the patterns — including how to use the surrounding words to pick the right one when two words sound alike. That's why a good system writes "I'll meet you there" and not "isle meat you their." It's reading the sentence, not just the sounds.
A second thing happens with the better tools: after the words are recognized, they get cleaned up — filler like "um" is removed, punctuation is added, and the small messes of natural speech are smoothed out. This matters more than it sounds, and we'll come back to it, because the difference between a raw transcript and a cleaned-up one is the difference between something you'd throw away and something you'd send.
What to expect on your first try
Your first attempt will feel slightly strange, and that's normal. Talking to a device to produce writing is a new motion, and like any new motion it's a little self-conscious at first. Two things will help.
First, start somewhere private — a quiet room, a parked car, a walk by yourself. The awkwardness of speaking out loud fades quickly, but it fades faster when no one's listening. Second, pick something genuinely low-stakes for the first go: a note to yourself, a reminder, a short message. Don't make your first dictation a high-pressure work email. You want room to fumble.
When you start, just talk. Say a sentence the way you'd say it out loud. Watch the words appear. You'll probably misspeak once or twice — and here's the most important habit in the whole guide.
The one rule that makes it work: don't edit while you talk
When you type, you fix things as you go without thinking about it. Resist doing that when you speak. If you say the wrong word, don't stop and fight it — just say the sentence again, or keep going and fix it later with your eyes. Trying to edit mid-sentence with your voice is what makes dictation feel clumsy and is the reason most people who give up give up.
So the rule is: speak when you're speaking, edit when you're editing. Get the words out first, messy and all. Then read it over at the end and tidy it up the way you would any draft. Keeping those two jobs separate is genuinely the difference between dictation feeling awkward and dictation feeling fast.
There's a reason this is hard at first. Drafting and editing are different mental modes, and a lifetime of typing has trained you to flip between them many times a second without noticing. Dictation doesn't let you flip that fast, which feels like a limitation but is actually doing you a favor — it forces the two jobs apart, and a lot of people find their drafts come out quicker precisely because they've stopped second-guessing every word as it lands. Trust the mess. The polish step is coming, and it's easier with a full draft in front of you than with a blank page you keep half-erasing.
Speak in natural chunks
Don't try to plan a whole paragraph before you open your mouth — you'll lose the end of it. And don't go word by word, which sounds stiff and kills the rhythm. The sweet spot is a phrase or a sentence at a time: think roughly what you want to say next, say it, then think the next bit. This is close to how you talk anyway, and it keeps things flowing.
You also mostly don't need to say the punctuation out loud. Good modern dictation figures out periods and commas from your natural pauses and the rise and fall of your voice. A small pause before a new thought helps it place the sentence breaks cleanly. Save the explicit instructions for things it can't guess, like starting a new paragraph.
Why the cleanup step matters so much
Here's the thing that surprises beginners. Natural speech, written down exactly, looks bad — full of "um," "you know," half-finished sentences, and little repetitions. That's not a flaw in your speaking; it's just what talking looks like on paper. If a tool hands you that raw, you'll think dictation doesn't work.
The good tools don't hand you that. They clean it up automatically, so what you see is closer to a real draft — punctuated, de-filler-ed, readable. When you're choosing where to start, this is the feature that matters most for a beginner, because it's the one that determines whether your first impression is "oh, this is great" or "never mind."
Give it a few days
Like any skill, this clicks through repetition, not reading. Use it for small things over three or four days — a few notes, a couple of messages, your thoughts after a meeting — holding to the one rule about not editing while you talk. Somewhere in there it stops feeling strange and starts feeling like the obvious way to get words down. That's the moment it becomes yours.
Quill is a good place to start from zero, because it handles the part beginners struggle with most. It transcribes and cleans your speech right on your iPhone or Mac, so the very first thing you see is a tidy draft rather than a raw transcript — and because all of it runs on the device, your early, fumbling practice sessions never leave your hands. Basic dictation is free with no word limit, so there's nothing to count while you're finding your feet. You can take your first try at quill.lumenlabs.works.