Google Docs Voice Typing: How to Use It (and What It Can't Do)

You've got a wrist that aches, a blank document, and a deadline. Somebody told you Google Docs can just take dictation. So you open the Tools menu, look for the thing, and either it's right there and you're off — or it isn't there at all and you spend twenty minutes wondering what you did wrong.

Usually you did nothing wrong. Google Docs voice typing is a genuinely good free tool that happens to be surrounded by bad instructions. Half the guides online still say it's Chrome-only. Half of them list voice commands that don't exist. And almost none of them tell you the four things that will actually annoy you a week in. So here's the whole thing: the setup, the commands worth memorizing, the browser situation as it stands today, and an honest map of where the feature stops.

Getting it running takes about ten seconds

There's no install, no add-on, no account upgrade. The feature is baked into Docs on the desktop web and it's free with any Google account, personal or Workspace.

1. Open a document in Google Docs on a desktop browser. It has to be a real Google Doc — if you opened a .docx file in Docs without converting it, the menu item won't be there. Use File → Save as Google Docs first.
2. Click Tools → Voice typing. Or hit Ctrl+Shift+S on Windows, Cmd+Shift+S on Mac. A small floating box with a microphone icon appears at the left edge of the page.
3. Pick your language from the dropdown at the top of that mic box. Do this before you start talking — switching mid-session drops what you were saying.
4. Click the microphone. The first time, Chrome (or Edge, or Safari) will ask for permission to use your mic. Click Allow. If you accidentally clicked Block once, months ago, the mic will just sit there doing nothing — fix it by clicking the padlock or tune icon in the address bar and flipping the microphone permission back on.
5. Talk normally. The icon turns red while it's listening. Don't shout, don't over-enunciate, don't go slow — Google's model was trained on ordinary speech and reads a natural pace better than a careful one.
6. Click the mic again to stop, or say "Stop listening." To pick back up, say "Resume" — it drops the cursor at the end of the paragraph and starts again.

One quiet detail worth knowing: the language dropdown holds more than 100 languages and regional accents, including thirteen flavors of English and twenty of Spanish. If you speak English with an Indian, Nigerian, Irish, or Filipino accent, picking that specific variant instead of plain "English (US)" is the single biggest accuracy improvement available to you, and it costs nothing.

The browser question, settled

This is the number one reason people search for help, and most of the internet is out of date on it.

Google's own support page now states that voice typing "works with the latest versions of" three browsers: Chrome, Edge, and Safari. Firefox is not on that list, and it's the one browser where the menu item genuinely will not work for you. So the old advice — "it's Chrome-only" — is no longer literally true.

But here's the sentence in Google's documentation that everyone skips, and it explains why Chrome still wins in practice: "When you turn on voice typing, your web browser controls the speech-to-text service." Google isn't doing the transcribing. Your browser is. So the same sentence, dictated into the same document, can come out differently in Safari than in Chrome — different engine, different vocabulary, different tolerance for background noise. Edge is Chromium under the hood but doesn't use Chrome's speech back end, and plenty of people report it loading the mic box happily, transcribing four words, and quitting.

Practical translation: if voice typing is misbehaving and you're not in Chrome, switch to Chrome before troubleshooting anything else. If it still misbehaves there, it's your mic permission, your document format, or your internet connection — in that order.

Tip: don't pause before punctuation.

The classic dictation failure is saying "…and that's the plan <pause> period." The pause tells the engine the sentence ended, so it treats "period" as a new word and types it out. Run punctuation straight on the end of the phrase, no gap — "and that's the plan period" — and it converts properly. The rule inverts for commands: those need a clear pause before and after, or "select all" just lands in your document as two words.

The commands worth actually memorizing

Google publishes well over a hundred voice commands — tables, columns, footnotes, text color, "dark purple 3." Almost nobody needs that. The useful set is maybe fifteen commands, and once they're muscle memory you can draft a full page without touching the keyboard.

Two hard constraints first, because they trip people up constantly. Voice commands only work in English — and both your account language and your document language have to be set to English. Dictation works in 100+ languages; commands don't. And commands don't work at all in Slides speaker notes.

Punctuation and breaks

"Period" · "Comma" · "Question mark" · "Exclamation point" · "New line" · "New paragraph"

Formatting

"Bold" · "Italicize" · "Underline" · "Strikethrough" · "Apply heading 1" (through 6) · "Apply normal text" · "Create bulleted list" · "Create numbered list" · "Increase font size" · "Highlight" · "Clear formatting"

Selecting text

"Select word" · "Select last word" · "Select last 3 words" · "Select line" · "Select paragraph" · "Select all" · "Select [word or phrase]" · "Deselect"

Editing

"Delete" (removes the word before the cursor) · "Delete last word" · "Delete [word or phrase]" · "Copy" · "Cut" · "Paste" · "Undo" · "Insert link" then say the URL · "Insert comment" then say the comment

Moving around

"Go to the end of the line" · "Go to the end of the paragraph" · "Go to the start of the document" · "Go to next heading" · "Go to next misspelling" · "Move down 2 lines" · "Scroll down"

The meta ones

"Stop listening" · "Resume" · "Resume with [word]" · "Voice commands list" (opens the full reference right inside your doc)

The powerful pattern is combining them. Say a sentence, then "select last 6 words", then "bold", then "go to the end of the line", and keep dictating. That's editing by voice, and it's what separates people who dictate for real from people who try it once and quit. Watch the little bubble on the mic box while you do it — it shows the last command it thinks it heard. If that bubble is wrong, say "Undo" immediately, before you dictate over the damage.

And one thing almost nobody uses: you can fix a typo with your mouse and keyboard without turning the mic off. Click into the mistake, retype it, click back, keep talking. The mic stays hot the whole time.

Slides gets a smaller version of the same tool

Google Slides has dictation too, but only for speaker notes — you can't voice-type onto a slide itself. Open a presentation, click Tools → Dictate speaker notes, and the same microphone box appears next to the notes panel.

It's a stripped-down build. Punctuation words still work. Voice commands do not — no "bold," no "select paragraph," no navigation. Which honestly suits the job: speaker notes are meant to be spoken anyway, so dictating them is one of the rare cases where talking to your computer produces exactly the register you want. If you're building a deck and narrating what you'd say on each slide, dictating the notes is faster than typing them and the result sounds more like you.

Tip: dictate in one pass, edit in another.

The most common reason dictation feels slow is that people try to speak and self-edit at the same time — talk, stop, stare, fix, restart. Speech and editing use different parts of your brain and fighting between them is exhausting. Get the whole messy draft out first, misheard words and all, then close the mic and clean it up with your hands. This works with any dictation tool on any platform, and it's usually the difference between "dictation is a gimmick" and "I'll never type a first draft again."

Where the tool runs out of road

Four limits. None of them are dealbreakers on their own; together they're why people who dictate seriously end up looking elsewhere.

It stops listening, silently. Click into another tab to check a fact, answer a Slack ping, glance at your calendar — the mic switches off. No warning, no chime. Long silences do the same thing. You come back, keep talking to a dead microphone for thirty seconds, and lose the paragraph. There's no autosave for words that never made it into the document.

It types exactly what you say. Every "um," every "so basically," every time you restart a sentence halfway through — it all lands in the document verbatim. That's not a bug, it's the design: this is transcription, not writing. Which means the time you save talking gets partly handed back cleaning up. A dictated page usually needs a real editing pass, and if you speak the way most people speak, that pass isn't short.

There's no custom vocabulary. You cannot teach it your CEO's surname, your product's name, your industry's jargon, or the acronym your team uses forty times a day. It'll guess, get it wrong, and get it wrong the same way tomorrow. For anyone whose writing is full of proper nouns — recruiters, salespeople, clinicians, lawyers — this is the limit that hurts most, because the words it fumbles are exactly the ones that matter.

It only lives inside one document. This is the real ceiling. Voice typing works in Google Docs and Slides speaker notes. That's it. The moment you switch to Gmail to send the thing you just dictated, or drop an update in Slack, or log a call note in HubSpot, you're back to typing with your fingers. You've learned a voice workflow that evaporates the second you leave one tab.

That last one is the gap tools like Laxis Voice Keyboard exist to fill — it sits underneath every app rather than inside one, so the same dictation habit works in Docs, your inbox, Slack and your CRM, and it cleans up the filler and false starts instead of transcribing them faithfully. It runs on Windows, macOS, iOS, Android and as a Chrome extension, and it supports a custom dictionary, which is the fix for the proper-noun problem. But be clear-eyed about the trade: Google's tool is free, already installed, and perfectly good for drafting inside a document. If that's all you need, you don't need anything else.

The bottom line

Here's what surprises people who stick with dictation past the first week: the hard part was never accuracy. Modern speech engines are good enough. The hard part is that speaking and writing are different skills, and most of us have spent twenty years learning to think through our fingers. The first three days of talking to a document feel wrong — stilted, self-conscious, slower than typing. Then something clicks and you find you can hold a paragraph in your head and say it, which is a thing you could always do out loud to a colleague but never onto a page. Whichever tool you use, give it more than one afternoon before you decide it isn't for you.

And if Google's version has you hitting its walls — Chrome-only, one document at a time, no cleanup pass — that's the case for a tool that works outside Docs too. We compared five of them across full workdays in our guide to the best dictation software in 2026.

Frequently asked questions

Does Google Docs voice typing work in Safari, Edge, or Firefox?

Google's own help page says voice typing works with the latest versions of Chrome, Edge, and Safari. Firefox is not on the list, so the Tools → Voice typing option won't appear or won't work there. Chrome is still the most reliable choice, because Google notes that your web browser — not Google — controls the speech-to-text service. Two browsers can transcribe the same sentence differently, and Chromium-based Edge uses a different speech back end from Chrome.

What's the keyboard shortcut for voice typing in Google Docs?

Ctrl+Shift+S on Windows and Cmd+Shift+S on Mac opens the voice typing microphone box. You can also get there from the menu with Tools → Voice typing. The shortcut opens the mic panel; you still click the microphone icon (or press the shortcut again) to start and stop listening.

Why aren't my voice commands working?

The most common reason is language. Google Docs voice commands are available only in English, and both your account language and your document language must be set to English — even though dictation itself supports more than 100 languages and regional accents. Commands also don't work at all in Google Slides speaker notes. If the language is right, pause briefly before and after each command, because a command spoken mid-sentence usually gets typed as literal text.

Can I use Google Docs voice typing on my phone?

No. Voice typing is a desktop browser feature — there's no Tools → Voice typing menu in the Google Docs mobile apps for iOS or Android. On a phone you dictate with your keyboard's built-in microphone (Gboard on Android, the iOS keyboard mic), which is a different engine with different behavior and none of the Docs voice commands.

Does Google Docs voice typing work offline?

No. It's browser-based and depends on a cloud speech-to-text service, so it needs an active internet connection. If your connection drops mid-sentence, the microphone usually stops without an obvious error and you lose whatever you were saying. It also won't run in a .docx file opened in Docs — the file has to be converted to native Google Docs format first.