Voice to text in Gmail, done right

Gmail has no built-in voice typing on desktop. Here is what works.

Open Gmail on a laptop, click Compose, look around the editor for a microphone. There isn't one. Google Docs and Slides have voice typing baked in (Chrome, Edge, or Safari), but Google's own support docs scope that feature to Docs and Slides. Gmail is not on the list. People have asked for a Gmail mic button in Google's support forums for years. It hasn't shipped.

The three honest paths on desktop: open Gmail on your phone instead — the system keyboard's microphone already works in any text field, including Gmail's compose. Install a Chrome extension that adds a microphone to the Gmail tab — Voice In, Voicy, or the bluntly-named 'Dictation for Gmail'. Or install a desktop app that intercepts a hotkey at the OS level and pastes transcribed text into whichever window has focus — Whisper, Willow Voice, Voicy's native build. The first works only in Chrome tabs. The third works anywhere a caret blinks.

The boring truth is these solve different problems. If you want voice only inside Gmail, an extension is enough. If you want it as a habit, for Gmail and the eight other apps you also write in, the extension hits its ceiling fast.

The fastest setup: a Chrome extension and a hotkey

If you want voice in Gmail and you're already in Chrome, this is the shortest path. Total time: about 3 minutes.

Prerequisites (one minute)

• Google Chrome installed (or a Chromium-based browser like Edge or Brave).
• A working microphone. Built-in mic is fine.
• OS-level mic permission granted. macOS: System Settings → Privacy & Security → Microphone. Windows: Settings → Privacy → Microphone.
• You are signed in to Gmail.

Pick one extension

The three with real userbases:

• Voice In by Dictanote, a Chrome-extension-only tool (also packaged as an Edge add-on). Works on 10,000+ websites in 100+ languages. Free tier and a paid Plus tier.
• Voicy, both a native app and a Chrome extension. The extension hooks into Gmail and 20,000+ other sites. Advertises 50+ languages and 99% accuracy.
• Dictation for Gmail, Gmail-specialised, adds a microphone icon next to the Send button. 60 languages, 100,000+ users, 3.5★ from 273 ratings.

Four-step install

Most people get this running in under three minutes. The friction isn't the setup. The friction is everything after, which we will get to.

On mobile, Gmail already does this, and that is fine

If you write most of your email on your phone, you do not need an app for this. You do not need an extension. You do not need to read the rest of this article. Total time to set up: zero. It's already there.

On Android, open Gmail, tap Compose, tap into the body, and look at the top row of the Gboard keyboard. There's a microphone icon. Tap it, speak, tap it again. Google's own docs phrase it as 'you can talk to write in most places where you can type with a keyboard.' They don't publish a clean language count, only that it doesn't work in every language Gboard supports. The major Romance and Germanic languages are covered.

On iPhone, open Gmail, tap Compose, tap the body, and look at the iOS keyboard. There's a microphone key next to the space bar. Tap it, speak, tap done. The iOS keyboard mic is system-level, so the same key works in Gmail, Messages, Notes, and Safari. (Gboard for iOS has its own mic button if you've installed it. Either works.) Apple doesn't publish a flat language count for iOS dictation; it tracks closely to macOS Dictation, which lists 47 regional variants.

Both are free. Both are built in. If you're a phone-first emailer, the rest of this article is for the days you're on a laptop. Otherwise, you're done.

Why the desktop extensions still get punctuation wrong

The Chrome extensions work, but their accuracy is the reason 'Dictation for Gmail' averages 3.5 stars from 273 reviews and not 4.7. The complaint pattern is consistent across all three. Once you know it, you stop being surprised.

Most browser-extension dictation tools are Web Speech API wrappers. They ask Chrome to do the speech-to-text, then paste the result. That API is fine for short utterances, fragile for anything longer. The specific things that break:

• Punctuation has to be spoken. 'Period' 'comma' 'question mark' 'new paragraph.' You think you'll learn it. You don't. Your brain does not want to say 'period' out loud at the end of every sentence. Most users dictate a flat run-on, then clean it up with the trackpad. Which defeats most of the point.
• The mic times out after silence. Pause for three seconds to think about phrasing, and the API stops listening. You start talking again, half a sentence is missing.
• Tab focus is sticky. Click into another Chrome tab and the mic stops. Switch to your calendar to check a date and the mic stops. Dictation lives inside one tab.
• Accents and proper nouns. Names get auto-corrected to whichever word the browser thinks is closer. Recipients of mine have been called Andre, Ankara, and once 'Ms. Andrescu' before I caught it.

Voicy claims its extension uses its own AI for higher accuracy, and that's the version I'd try first if I were committing to the extension path. But none solve the punctuation-by-voice and timeout problems. Both are baked into the assumption that browser dictation is a 30-second action, not a writing tool.

The same browser-extension ceiling applies to voice typing in Google Docs — handy inside the tab, but it never follows your cursor into Gmail or anywhere else.

What I use for Gmail, and when I do not bother

I built Whisper because I'd burned out on Chrome extensions for the reasons above. It is a Tauri desktop app for Mac (Apple Silicon) and Windows. No Intel Mac, no Linux. It registers a hotkey at the OS level. Hold, speak, release. The text is transcribed and pasted into whichever window has focus.

Default hotkey: Ctrl+Space on Windows, modifier-only Command + Option on macOS (either Command, either Option, held to record, released to stop). Both customisable in Settings.

The piece that changed how I work: it doesn't care which app has focus. Same hotkey in Gmail compose, in Slack, in Notion, in VS Code, in the terminal. The Rust paste module writes to the OS clipboard and injects the native Cmd+V or Ctrl+V shortcut into whatever window I'm looking at. macOS asks for Accessibility permission the first time. After that, it's invisible.

For Gmail, this matters because of context-switching. I'll start composing a reply, alt-tab to my calendar, alt-tab back. With a Chrome extension, the mic state is gone. With a desktop hotkey, I pick up where I left off.

You pick the transcription engine. Three paths, presented honestly, no default:

• Cloud (OpenAI BYOK). gpt-4o-mini-transcribe by default, gpt-4o-transcribe for the higher-accuracy tier. You bring your own OpenAI key. Best quality, web search built in, needs a connection.
• Local Parakeet. NVIDIA's TDT model, around 600 MB, 5–10× faster than Whisper on CPU. English plus 24 European languages. Offline. No Asian languages, no translate-to-English.
• Local Whisper. Base (around 140 MB) up to Large v3 (around 3 GB). Multilingual variants cover 99 languages. The .en variants are English-only. Slower than Parakeet, but supports translation and finer per-recording control.

For Gmail replies in English, Parakeet local is what I leave it on. Cloud takes over when I need a longer email with web search. The AI keyword trigger is 'Hey whisper' followed by the instruction.

When I don't bother: a one-line 'thanks, will do' reply. The hotkey latency beats typing only past ten words or so. The point of dictation is the 80-to-200-word emails where speaking is faster than typing.

When Gboard or iOS dictation is the right answer

If 90% of your Gmail happens on your phone, install nothing. Gboard and iOS dictation are free, system-level, and built for short-to-medium messages. The mic icon on your keyboard handles a 100-word reply in 40 seconds.

If you're on a Mac and your needs are 30-second voice memos in a text field, Apple Dictation works for short notes. macOS Tahoe ships with 47 languages and regional variants. Activated by the Microphone key, a configurable shortcut, or Edit → Start Dictation. It stops after 30 seconds of silence, which makes it great for 'yes, reply to confirm' and bad for a paragraph that needs a pause to think.

If you write the bulk of your emails on a desktop or laptop, in multiple apps that aren't all browser tabs, that's where a system-level app starts to matter. We make one. So do Willow Voice and Voicy's native build. Try a few before settling — the same system-wide hotkey also lets you dictate into Notion.

Voice typing in Gmail should have been a Google feature ten years ago, and somehow still isn't. Until it is, pick the path that fits your day. Phone keyboard if you're on the phone, Chrome extension if you live inside one tab, desktop app if you write everywhere your cursor goes. Whichever you pick, the goal is the same. Stop typing the 100-word emails.