← Guides · atomAI Scribe · Dictation

Voice typing on Windows: Win+H, Voice Access or a dictation app?

Last updated: October 12, 2026

Windows 11 ships with two speech features, and they are easy to mix up. One is for typing with your voice, the other is for operating the whole computer by voice. Both are free and already on your PC. This guide explains what each one does, where they run out of road if you dictate every day, and when a separate dictation app earns its place.

Voice typing (Win + H)

Press Windows + H in any text field and a small bar with a microphone appears. Speak, and the text lands at your cursor. It is the quickest way to try dictation on Windows, and for a sentence here and there it is genuinely fine.

The part most people do not know: voice typing does not recognise your speech on the PC. It sends the audio to Microsoft’s online speech service and gets the text back. That is why it needs an internet connection, and why it only works once online speech recognition is switched on under Settings → Privacy & security → Speech.

Voice Access

Voice Access is the other built-in feature (Settings → Accessibility → Speech). It is designed to operate the whole computer by voice — open apps, click buttons, scroll — and it can dictate as well. Its recognition runs on the device, so it works without a connection.

The trade-off is its focus. It listens for commands as much as for text, it takes some learning, and it is not available in every language. For people who cannot use a keyboard or a mouse it is an important tool. As a pure writing tool it is more than most people need.

Where the built-in options run out

If you dictate a few minutes a day you may never notice. If you write emails, notes and documents by voice all day, you will run into these:

• Your own words. Names, product names and jargon come out wrong until a tool lets you teach it how you spell them.
• The connection. With Win + H, no network means no dictation — on a train, on a plane, or behind a strict company network.
• Where the audio goes. If you dictate client, patient or confidential text, “it goes to a cloud service” is a question your employer or data-protection officer will ask.
• Structure. A new paragraph, taking back the last sentence, sending text to the clipboard — small things you want to do without reaching for the mouse.

Side by side

As of September 2026. Windows features change with updates; check the settings on your own PC.

When a dictation app is worth it

Short version: when dictation becomes a habit. The built-in tools are made for everyone; a dictation app is made for people who write by voice every day. It is worth paying for once it saves you more time than it costs, and if you write for a living, that happens quickly.

How atomAI Scribe does it

• Recognition on your PC. Scribe does the recognition itself, on an ordinary processor — no graphics card needed. Your audio is never uploaded.
• One hotkey, every program. Press Ctrl + Alt + D (you can change it), speak, press again. The text lands at your cursor in any Windows program with a text field.
• Punctuation and capitals are set automatically — you do not say “comma”.
• Four spoken commands at the end of what you say: “Scribe, new paragraph”, “Scribe, scratch that”, “Scribe, copy that” and “Scribe, cancel”.
• Your own dictionary and text shortcuts from the Standard tier. Optional settings remove filler words and resolve spoken corrections (“Monday — no, Friday”).
• Twenty-three languages on every tier.

Good to know before you try it

• Scribe runs on Windows 10 and 11 (64-bit) only.
• Setup downloads the speech model once, about 640 MB. After that, dictation works offline.
• Windows shows a SmartScreen notice on first run, because the installer is not signed with a certificate yet. Click “More info”, then “Run anyway”.
• The app installs for free; dictation unlocks when you buy it — once, or as a yearly plan.