Google Docs Voice Typing on Mac: Limits and the Fix

Google Docs voice typing on Mac runs in Chrome, Edge and Safari, but only in English and only in that tab. Here is what it cannot do, and the way around it.

Google Docs has voice typing built in, under Tools > Voice typing, and it covers more browsers than its reputation suggests: Google lists the latest Chrome, Edge and Safari. The catch is the narrowness. You speak the punctuation out loud every time, saying “period”, “comma”, “new paragraph”. Those voice commands are English only, and Google requires both your account language and your document language to be English. Google also says your browser, not Google, controls how your speech is processed. And none of it helps anywhere else on your Mac.

Dictate at the system level, not inside the page

Let macOS listen instead. Dictato sits in the menu bar: put the cursor in the document, press the shortcut, speak, press again. The finished text is copied to the clipboard and pasted with a simulated Command+V at the cursor, the same way in any browser. Your clipboard content is restored afterwards. Delivery needs macOS Accessibility permission, recording needs microphone permission, and the same route works in any app on your Mac.

One rule matters: the text goes back to the window that was focused when you started dictating. Switch apps mid-dictation and it does not paste blind: it sounds an error and opens a recovery window instead. Every dictation goes to local history before the paste, kept 30 days by default, your safety net if a page quietly swallows it.

Give docs.google.com its own profile

A per-app profile can be scoped to a browser tab address, so document writing gets its own tone and formatting. Patterns are hosts only: docs.google.com or *.google.com, never a path. Address reading works in Safari, Chrome, Brave, Edge, Arc, Vivaldi and Opera; in Firefox a rule covers the whole browser. The first time a site rule needs to read the address, macOS asks for Automation permission; decline and it falls back to the browser-wide rule. Your first profile using the text clean-up downloads a local AI model once, around 3.4 GB. The per-app profiles walkthrough covers the setup.

Long documents do not need a timer

There is no silence timeout: recording runs until you stop it, so pausing to think does not end it. A single recording tops out at two hours, then stops and transcribes normally, nothing discarded.

Formatting arrives in two layers. Spoken numbers, dates, times, amounts and percentages become written form with nothing to switch on, in English and French today. Sentence punctuation, commas, capitalisation and paragraph breaks come from the on-device clean-up, which only runs once you enable Auto-correct or set up a per-app profile. Neither is on out of the box. Paragraph breaks follow where the subject changes, never where you paused. That is what removes the need for saying the punctuation out loud while you draft.

Recording, transcription and clean-up make no network calls, so your audio and text stay on the Mac, apart from that one-time model download, licence and trial checks, and feedback you send.

A method that produces a draft

Outline first, headings only. Dictate section by section and resist editing while you speak: consecutive dictations chain with correct spacing. Correct with the keyboard at the end; undo is your editor’s own.

It is a 7-day trial with no account and no card, then 19.99 EUR once, for life.