Most of the time spent on email isn't reading it. It's writing the replies — the ten short ones that each need a sentence or two, and the two long ones you keep putting off. Email is basically talking. You already know what you want to say to the person. So say it.
Dictating email on a Mac is one of the cleaner wins for voice to text, because the writing is short, conversational, and there's a lot of it. This is how to set it up across Gmail, Apple Mail, and Outlook, and where it stops being worth it.
Dictating it vs. letting AI write it
There are two different things being sold as "AI email" right now, and they're not the same.
One is the wave of tools that compose the email for you — you give a one-line prompt and Apple's Writing Tools, Gemini in Gmail, or Copilot in Outlook generates a full message. That's fast, but the output reads like it was generated, because it was. The recipient can usually tell. And you still have to read it carefully to make sure it says what you meant.
The other is dictation: you speak the actual message and it lands as text. It's still your words, your tone, the specific thing you wanted to say to this specific person — just spoken instead of typed. For anything where the relationship matters, that difference shows. Dictation keeps you in the message; auto-compose takes you out of it.
This post is about the second one.
Why email suits voice
The messages are short. A two-sentence reply is roughly ten seconds of speech. Typing it means clicking in, and the ten-second reply somehow takes a minute.
There's volume. The speed gap between talking and typing (about 130–150 words per minute versus 40–60) compounds across thirty replies a day, not one.
It's conversational by nature. Email is closer to how you talk than most writing is, so the spoken version usually needs less fixing than, say, a dictated report would.
The one setup that covers every mail client
You don't need a Gmail plugin, an Outlook add-in, and an Apple Mail extension. A system-wide dictation app types wherever your cursor is, so the same setup works everywhere:
- Apple Mail — click into the reply body, dictate, done.
- Gmail in the browser (Safari, Chrome, Arc, Firefox) — click into the compose box and dictate. Works the same in the Gmail web app as in a native client, because the app pastes text at the cursor, not through Gmail's API.
- Outlook — the desktop app or Outlook on the web, same story. Click into the body, talk.
- Spark, Superhuman, HEY, Missive — anything with a text field. If you can type into it, you can dictate into it.
That's the advantage of dictation that works at the OS level over a feature baked into one mail client: your inbox setup can change and the voice workflow doesn't.
What the built-in options give you
Apple Dictation is free and on-device, and for a quick one-line reply it's genuinely fine. Toggle it on in System Settings and it types into Mail or a browser compose box. The friction shows up on longer replies — it's built for short bursts, and it transcribes filler words and false starts literally, so a dictated paragraph reads like a transcript of you thinking. For where that line falls, we wrote a whole piece on Apple Dictation versus a dedicated app.
The gap a dedicated tool fills for email specifically is cleanup. You want the sent message to read like you wrote it deliberately, not like you dictated it. That's a cleanup pass, and Apple Dictation doesn't have one.
Setting up Vext for email
Vext is a $49 one-time, fully local dictation app — here's the email configuration:
- Install:
brew install muvon/tap/vext - Grant Microphone and Accessibility permissions
- Keep the default push-to-talk hotkey — for short replies, hold-talk-release is the right rhythm
- Turn on Enhance (default Gemma 3 4B) so "um yeah so basically can we push the call to Thursday" becomes "Yes — can we push the call to Thursday?"
- Leave YOLO Mode off. It auto-submits after pasting, which is great for AI prompts and exactly wrong for email. You always want to read an email before it sends.
The workflow that clears an inbox fast: open the reply, hold the hotkey, say the message the way you'd say it out loud, release. The cleaned-up text lands in the compose box. Read it. Send. For the ten short replies, this turns twenty minutes into five.
The etiquette and accuracy notes
A few things worth getting right, because email goes to other people.
- Always read before sending. Dictation is accurate, not perfect. A misheard name or a wrong number in a reply is worse than a typo you'd have caught while typing. This is the one place to never skip the review.
- Spell out names and addresses, or type them. Proper nouns and email addresses are where any speech engine struggles most. For anything that has to be exact, type it.
- Mind the room. Dictating email in an open office means your colleagues hear your replies. Fine for "sounds good, thanks," awkward for a sensitive HR note. Read the room, literally.
- Sensitive content stays private with a local tool. If the reply involves anything confidential — legal, financial, personnel — a fully on-device app matters, because the audio never leaves your Mac. Cloud dictation services route it through their servers.
Where voice doesn't help
The long, careful email. The one to a client where every sentence is weighed — dictate a first pass if you want, but you'll edit it as heavily as anything else. Voice speeds up the draft, not the deliberation.
Anything with a lot of exact detail. Invoice numbers, URLs, code snippets, precise figures. Faster to type than to dictate and then fix.
When you're not sure what to say. Voice is for getting a message you already have out of your head. If you're still deciding what the message is, talking to yourself isn't much faster than typing to yourself.
For the rest — the daily stream of short, conversational, already-decided replies that eat your morning — talking through them is meaningfully faster, and the results still sound like you.
Download Vext — 100 free dictations, no account, no card. Try it on tomorrow's inbox and see how much of it you can just say. For a wider look at the options, our best voice to text apps for Mac roundup covers the rest of the field.