How to Dictate in Any Windows App — Word, Outlook, Chrome and the Rest

Most dictation advice on Windows is written as if you only ever dictate into one program. Real work is not like that: a paragraph in Word, a reply in Outlook, a comment in a browser, a message in Teams — and each one has its own quirks.

Here is the practical version.

Start with what is already installed

Press Win+H anywhere in Windows. A small toolbar appears, and whatever you say is typed into the field your cursor is in. It works in Word, Outlook, Chrome, most chat apps — anywhere that accepts normal keyboard input.

Do this before you buy anything. If your needs are short notes and quick replies, you may not need another tool at all, and there is no sense in paying for one.

Where it stops is quality. Win+H writes down what you said, exactly as you said it — including the “um”, the false start, and the sentence you began twice. It will not tidy grammar, adjust the tone of a message, or learn how to spell a name it keeps getting wrong.

The one thing that decides whether dictation sticks

It is not accuracy. It is where the text lands.

Any tool that makes you dictate into its own window and then copy the result into your document will be abandoned within a week — the copying is the friction, and it never goes away. What works is a tool that types into whatever window has focus, the way a keyboard does.

That single behaviour is what makes dictation feel like typing rather than like a workflow.

App by app, what to expect

Word. Works well with everything. Word has its own dictate button too, but there is no reason to switch between two systems if you already dictate elsewhere.

Outlook and Gmail. Both are just text fields. Dictation that types into the focused window works in either without configuration. The thing that actually matters in email is not transcription — it is that dictated sentences read as spoken, which is why a tool that can rewrite what you said into a professional register saves the most time here.

Browsers. Any text box, any site. To the browser, typed text is typed text.

Chat apps. Watch the Enter key: many chat apps send a message on Enter, so a dictated line break can post half a sentence. Dictate the message, then read it, then send it.

Editors and IDEs. Fine for comments, commit messages and documentation. Not for code — no dictation tool handles punctuation-dense syntax well, and anyone who tells you otherwise has not tried to dictate a regular expression.

When dictation suddenly stops working

Two causes cover most of it.

Administrator mismatch. If the target application runs as administrator and your dictation tool does not, Windows will block the simulated keystrokes. The fix is to run both at the same level.

Non-standard text fields. Remote-desktop sessions, virtual machines and a few older applications do not expose normal Windows text controls. Dictation into those is unreliable regardless of the tool.

Stop the second pass

The part people underestimate is the editing afterwards. Raw transcription always needs a clean-up pass, and that pass is done with the keyboard — which cancels out much of what you saved.

This is where the current generation of tools differs from the old ones. Instead of only transcribing, they can clean the grammar and drop the filler words, or rewrite what you said as a professional message, before the text is typed. The goal is that what lands in the document is what you actually wanted to write.

That is what BeltoVox does on Windows: it types into whatever app has focus, and you choose what happens to the text on the way — raw, cleaned up, or rewritten. It is $59 once rather than a monthly fee, and it runs on your own transcription API key.

If you are only dictating occasionally, use Win+H. If you are doing it every day and then fixing the result by hand, that second pass is the thing worth removing.