What is online speech recognition in Windows?

Online speech recognition is the setting in Windows 11 that decides where your voice is turned into text. When it is on, the audio you dictate goes to Microsoft's cloud ¹, and the text comes back. When it is off, Windows uses device-based recognition, and only device-based features stay available ². Voice typing is the feature Microsoft names as depending on it ³.

Online recognition uses Microsoft's cloud for more accurate results

Microsoft describes two kinds of speech recognition in Windows. Device-based recognition processes your voice locally, and no voice data is sent to Microsoft ⁴. Online speech recognition uses Microsoft's cloud, and Microsoft says that is for more accurate recognition ¹.

The trade-off is in the data path. Microsoft says voice data is sent to Microsoft only to provide the service and create text transcriptions, and that it does not store, sample or listen to voice recordings without your permission ⁵. So the audio is sent to produce the transcript. Whether Microsoft keeps any of it is a separate choice, covered below.

Microsoft presents online recognition as one switch in Windows, with no per-app choice on the speech privacy page we read.

Voice typing is the feature that uses the cloud

Microsoft says that voice typing "converts speech to text using online speech recognition" ³. So when the switch is on, voice typing sends your audio to the cloud, and when it is off, voice typing stops.

When you turn the switch off, Microsoft says only device-based speech features remain, and it gives Narrator as the example ². For dictation that must stay on the PC, see our question on whether Voice Access sends your voice to Microsoft.

Voice clips are a separate setting

Turning on online speech recognition does not mean Microsoft keeps your recordings. Keeping them is a different choice. Microsoft says you can use voice typing without contributing voice clips, and that contributing them is optional ⁶. If you do contribute, Microsoft says it keeps the clips for up to two years, to be sampled and listened to ⁷.

This matters for the question you are asking. "Online" describes where recognition runs, while the clip setting describes what Microsoft may keep afterwards. Check both.

It depends: which features use the switch

Microsoft's speech privacy article names voice typing as the online feature and Narrator as a device-only feature. It does not give a full list of every feature that uses online recognition. So if a feature is not named, do not assume its path. Test it, or ask your IT team if the PC is managed.

Work and school accounts are a special case. Microsoft says sharing voice data is not available on a work or school account, but voice typing still works ⁸. Audio still goes to Microsoft for voice typing on those accounts ⁵. We did not verify whether an administrator can turn off online speech recognition by policy, and Microsoft's pages we read do not say.

What to do

1. Open Start, then select Settings.
2. Select Privacy & security, then select Speech.
3. Set Online speech recognition to Off if you do not want voice data sent to Microsoft. Voice typing will stop until you turn it back on.
4. Open voice typing, select Settings, and check that voice clip contribution is off.

These steps are for Windows 11. Menu names can change with an update, so check them against your version.

If you want dictation that runs on your PC, Yap, which we make, is an on-device dictation app for Mac and Windows. Compare it with the built-in tools on the same questions. For the full picture, read our guide to Windows dictation privacy, or the wider overview of Windows voice typing. Related questions include whether Windows dictation is HIPAA compliant and whether voice typing works in every app.