VibeVoice

Voice typing in Word

Press your hotkey and speak; punctuated text arrives in the document at the cursor. Nothing is installed into Word, no template changes, no add-in to approve — and the same hotkey works in your mail, your browser and everything else on the machine.

A small glowing orb sending a single ribbon of speech into a stack of glass pages, the nearest one filling with lines of text.

Word already has a Dictate button. Why use anything else?

Dictate is genuinely good and it is included with a Microsoft 365 subscription, so if Word is where all your writing happens, use it. The case for this is the sentence after: the same hotkey also types into Outlook, a browser form, a chat window and a PDF form, so you learn one thing instead of one per application — and it does not need a 365 subscription.

We have not benchmarked Dictate and make no claim about which transcribes better. The difference on offer is where you can use it, not how the words come out.

One hotkey, wherever the cursor is

The document, the email replying to it, the browser tab you checked a figure in. Dictation follows the cursor rather than the program, so there is nothing to learn twice.

Nothing installed into Word

No add-in, no macro and no template to modify, so there is nothing installed for an IT department to approve. Word does not know it is there; it sees keystrokes.

Letters and long drafts

Formulaic writing is where speaking gains most — a report, a covering letter, the third version of a proposal. Recordings go in as files too, up to 10GB on Ultra.

How it works with Word

Three steps, none of them inside Word.

  1. Install the desktop client

    A native binary for Windows, macOS or Linux, linked to your account once with a device code. It runs in the tray; Word never sees it and there is no add-in to approve.

  2. Put the cursor in the document

    The body, a heading, a table cell, a comment, the subject line of the mail you are replying to. The client types into whatever is focused.

  3. Press your hotkey and speak

    Punctuated text arrives at the cursor as you go. Read it back, carry on typing if you would rather. Nothing is saved or sent for you.

What you dictate, and what stays on the keyboard

You dictate the prose. Headings, tables, numbered lists, tracked changes and citation fields stay on the keyboard, because the client types words into the field in front of it and takes no commands. That sounds like a large exception and is a small one: in a letter, a report or a witness statement, the body text is most of the words.

It is also the part that costs the most. Formatting a document is fiddly but brief; drafting it is the hour. Moving the hour off the keyboard is the whole of what this does — and if the reason you are reading this is that the hour hurts, the page on dictation and hand pain is the more careful one.

What it does not do

  • No formatting by voice. Word’s own Dictate takes commands like “bold that” and “new heading”; this types words and punctuation, and the formatting stays on the keyboard.
  • No add-in, which is the point above stated as a limitation: it cannot see your document, so it cannot move the cursor, select a paragraph or read anything back to you.
  • Windows and macOS for Word itself, because that is where Word runs. The client runs on Linux too, where the browser version of Word is a text field like any other.
  • No offline mode. Audio is transcribed on our servers, so a train with no signal is a train where this does not work.

Common questions

Do I need Microsoft 365 for this?+

No. Word’s built-in Dictate is part of a Microsoft 365 subscription; this is separate software that types into whatever window has focus, so it works with a perpetual Word licence, an older version, or no Word at all.

Does it work in Word on a Mac?+

Yes, with the same hotkey. macOS asks for Accessibility permission the first time, because it will not let any application send keystrokes to another one without it. That is granted once, in System Settings.

What about Word on the web?+

That is a browser tab with a text field in it, so it behaves like any other page: the text lands where the cursor is. The exceptions across the board are programs that deliberately reject synthetic input — some games, some remote-desktop and virtual-desktop sessions, and some hardened enterprise clients.

Can I dictate punctuation?+

You do not have to. Full stops, commas, question marks and sentence breaks are applied from the way you speak, which is the part that saves the most time on a long document. There is no spoken command vocabulary at all — nothing to learn, and nothing you have to remember not to say.

Can I transcribe a recorded meeting into a document?+

Yes, as a file rather than live: upload the recording, get the transcript, paste it where you want it. Speaker separation is available on Ultra for a conversation with several people in it.

The same client types into Google Docs, Gmail and Notion. Drafting for a court or a clinic? Dictation for lawyers and for doctors go further.

Last updated .