How to Manage Advanced Speech Recognition Voice Models in Windows 11

Navigating to Speech Settings in Windows

Every task in this guide starts in the same place, the Speech page in Settings. Open it once and you can reuse the same route for each section below.

Two related Windows 11 features use these speech models, and it helps to know which is which before you change anything. Voice typing opens with Win+H in any text box and types what you say. Voice access, found under Accessibility › Speech in Settings, lets you control the whole PC by voice.

Voice access needs Windows 11 version 22H2 or later, with current updates installed. In newer builds, both features can use speech models stored on your own computer instead of sending audio online. On an older build, voice typing still works, but Voice access may be missing from Settings. To check your version, open Settings › System › About

1. Press Win+I to open Settings. The Settings ⚙️ window opens with a list of categories on the left.
2. Click Time & language in the left-hand navigation pane. The page on the right changes to show date, region, and language options.
3. Click Speech. The voice recognition and text-to-speech configuration page opens, with the language, voice, and microphone options in one place.

If you cannot find Speech, check that Windows 11 is fully updated. Some options on this page appear or move slightly between builds, so a label on your screen may differ a little from the ones used here.

Configuring Speech Language and Voices

The Speech language setting decides which language Windows listens for when you use voice typing, and which language it uses when it reads text aloud. If this does not match the language you actually speak, you get wrong words or no output at all. Choosing the right language and region lets Windows understand your accent and vocabulary correctly.

Before you start, know that a speech language pack must be installed for dictation and for Voice access to work. A language pack is a set of files that gives Windows the vocabulary and speech model for one language. Downloading it needs an internet connection. If the pack fails to install, restart the PC and try the download again.

1. Open Settings, go to Time & language, and click Speech.
2. Find the Speech language dropdown menu near the top of the page.
3. Click the dropdown menu and select the primary language you speak from the installed packs. Pick the regional variant that fits you, for example English (United Kingdom) rather than English (United States), because the models differ in accent handling.
4. If your language is missing from the list, click the Add a language button, search for your language pack, and follow the prompts to download it. The download needs an internet connection, and a large pack can take a few minutes.
5. Choose a specific voice profile from the Voices dropdown if you want to change the accent or pitch of text-to-speech features. The available voices depend on the language packs you have installed.

To undo the change, return to the same dropdown and select your previous language. You do not normally need to restart, but if an app still listens in the old language, close it and reopen it.

Managing Microphone and Voice Training Options

The microphone setup on the Speech page lets Windows check your input device. This helps when dictation is full of errors or voice commands stop working. Run it whenever you switch to a new headset or microphone.

1. Open Settings, go to Time & language, and click Speech.
2. Scroll down to the Microphone section, where your input devices are configured.
3. Click the Get started button under the microphone setup heading if your system fails to pick up voice commands. A setup window opens.
4. Follow the on-screen prompts and read the short sentence aloud so Windows can calibrate your voice input. Speak at your normal volume and distance from the microphone, in a quiet room if you can.

If the setup cannot hear you at all, the wrong device is probably selected. Check the troubleshooting section at the end of this guide.

Adjusting Text-to-Speech Speed and Audio Output

You can change how fast Windows reads text aloud so it matches how you like to listen. This only affects text-to-speech, not how Windows recognizes what you say.

1. Open Settings, go to Time & language, and click Speech.
2. Find the Text-to-speech section further down the page.
3. Click and drag the Voice speed slider to the left to slow down the reading rate, or to the right to speed it up.
4. Click the Preview voice button to hear a sample at the new speed. If it is too fast or too slow, adjust the slider and preview again.

There is no separate save button. The setting applies as soon as you move the slider, and you can drag it back at any time to undo it.

Controlling Online Speech Privacy via Settings and Registry

Turning this off keeps more of your voice data on your own computer. The trade-off is that features that rely on online recognition may become less accurate or stop working.

The Online speech recognition switch in Settings does not need administrator rights. Editing the registry key may need them if the policy is machine-wide. The setting lives under Privacy & security, not under Time & language.

1. Open Settings, click Privacy & security, and select Speech.
2. Toggle the Online speech recognition switch to Off to stop sending voice data to Microsoft servers. The switch should now show Off.

To turn it back on, return to the same page and set the switch to On.

Before you edit anything, click File, then Export, and save a backup of the key so you can restore it if something goes wrong.

1. Press Windows key + R, type regedit, and press Enter to open Registry Editor. Click Yes if User Account Control asks for permission.
2. Navigate to the following path in the left sidebar: HKEY_CURRENT_USER\Software\Microsoft\Speech_OneCore\Settings\OnlineSpeechPrivacy
3. Double-click the target value in the right pane to modify its data. The Edit dialog opens with the current value.
4. Change the data, click OK, and close the Registry Editor.

Sign out and back in, or restart the PC, if the Settings page does not show the new value. To undo the change, set the value back to what it was or import the backup you exported.

Troubleshooting Common Speech Recognition Failures

When voice typing stops working or apps cannot hear your microphone, the cause is usually the wrong input device or a mismatched language. Work through these checks in order.

1. Right-click the Speaker icon in your taskbar and select Sound settings.
2. Scroll down to the Input section and confirm that your correct microphone is selected as the default recording device. Speak and watch the input level bar to see if Windows picks up sound.
3. Open Settings, go to Time & language, and click Speech to re-select your primary language pack.
4. Restart your computer to clear temporary audio cache files and reload the speech recognition services.

If the microphone still does not work, check that microphone access is allowed under Privacy & security in Settings, and that the microphone is not muted by a hardware switch or key on your headset or laptop.

If dictation or Voice access still fails after that, confirm the speech language pack for your language is installed. A pack that did not finish downloading, or failed to install, can leave both features silent. Connect to the internet, add the pack again, and restart the PC if the install fails.

What to do next

You should now see your updated language and voice configurations active across your system apps. To go further, open the neighboring Time & language pages and set up regional formatting and keyboard layouts to customize your typing experience.