How to Set Up Custom Speech Recognition Profiles in Windows 11 Time and Language Speech Settings

Navigating to Time & Language Speech Settings

The speech settings page is where you manage your spoken language, your microphone, and related voice options. Every later step in this guide starts here or on a nearby page, so it helps to know the route.

1. Press Win+I to open Settings. The Settings⚙️ window appears with a list of categories on the left.

2. Click Time & language in the left-hand sidebar. The page lists date, language and speech options.

3. Click Speech. You should now see the speech configuration screen, where you manage spoken input, microphone selection, and voice options.

The layout of this page varies a little between Windows 11 versions. If a section is in a different place on your PC, look for it by name rather than by position.

Windows 11 also has Voice access, found under Settings › Accessibility › Speech. It is the modern replacement for the legacy Speech Recognition tools. It runs on your device once its language files are downloaded, lets you control the whole PC by voice, and supports a custom vocabulary. It needs Windows 11 version 22H2 or later, so if you cannot find it, check Windows Update.

Managing Speech Languages and Regional Packs

The speech engine needs to know which language and regional variant you speak. The right choice matters most if you have an accent or a regional dialect. A US English model, for example, will misread many words spoken in other English variants.

1. Open Settings, go to Time & language, and select Speech. The speech page opens.

2. Find the Choose the language you speak with your device drop-down menu.
3. Open the menu and select your native language or regional dialect. The setting takes effect straight away, so there is no Save button.
4. If your language or dialect is not in the list, you need to install its language pack. Go back to Time & language and select Language & region. The list of installed languages appears.

5. Click Add a language and search for the language you want.
6. Select the language and click Next. A screen of optional language features appears.
7. Check the box for Speech recognition, then click Install. The download needs an internet connection and can take a few minutes.
8. When the install finishes, return to Speech and pick the new language from the drop-down menu.

If Speech recognition is not offered for a language, that language has no speech support in Windows. In that case, choose the closest supported regional variant.

To undo the change, switch the drop-down back to your previous language. You can also remove the added language from Language & region. A restart is not normally needed.

Setting Up and Testing Your Microphone

1. Open Settings, go to Time & language, and select Speech. The speech page opens.

2. Find the Microphone section near the top of the page.
3. Click the Get started button under the microphone setup option. A short test window opens.
4. Read the sentence aloud at a normal volume and follow the on-screen prompts. Watch the audio meter. It should move clearly when you speak, but it should not hit the maximum and stay there. Clipping and distortion lower recognition accuracy.
5. If Windows picked the wrong device, set the right one as your default input. Open the Windows Sound settings (under System in Settings) and choose your headset or standalone microphone as the input device. The sound options in Control Panel🎛️ offer the same choice.
6. Repeat the test after changing the device to confirm the meter responds.

If Get started is greyed out or the meter stays flat, check that the microphone is plugged in and not muted. Also check that microphone access is allowed under Privacy & security › Microphone. A USB headset or external microphone usually gives better results than a laptop’s built-in one.

Training Your PC to Better Recognize Your Voice

This part adapts the legacy speech engine to you. The training wizard in Control Panel records you reading sample sentences and adjusts the recognition profile to your pitch, pace and pronunciation. Do it in a quiet room, with the same microphone you will use for dictation.

1. Press the Win, type Control Panel🎛️ in the search🔍 bar, and press Enter. Control Panel🎛️ opens.
2. Click Ease of Access, then select Speech Recognition.
3. Click Train your computer to better understand you. The voice training wizard opens.
4. Read each sentence aloud in your normal speaking voice and at your normal pace. Do not slow down or over-enunciate, or the profile will be tuned to a voice you do not actually use.
5. Continue until the wizard says training is complete. The recognizer uses the new data from then on.

If Speech Recognition has never been set up on the PC, Windows may first start a short setup wizard for the microphone. Complete it, then run the training again.

You can run the training session more than once. Each round tends to improve accuracy, especially if you use technical terms.

Configuring Advanced Speech Options

The advanced options control how the legacy speech tools behave day to day. They also let you manage the recognition profile Windows keeps for you. Review these settings once so nothing runs, or collects data, that you do not want.

1. Open Control Panel🎛️ from the Windows search menu.
2. Click Ease of Access and select Speech Recognition.
3. Click Advanced speech options in the left-hand menu. The Speech Recognition Properties window opens.
4. Check Review documents and mail to improve speech recognition if you want the system to look at your writing so it can predict words better in context. Leave it unchecked if you would rather it did not scan your documents.
5. Clear the startup option if you do not want voice tools to launch automatically every time Windows boots.
6. Click Apply, then OK, to save the changes.

The window also lists the recognition profiles stored on the PC. On most builds, a separate profile for a second person or a different microphone is created with the New button, and an unwanted one is removed with Delete.

The exact wording of these options can differ a little depending on your Windows build. If you change your mind, return to the same window and reverse the checkboxes.

Configuring Privacy Settings for Speech and Voice Activation

Cloud-based speech recognition and voice typing only work if Windows is allowed to send your speech to Microsoft for processing. These switches are in Privacy & security. If they are off, voice typing may fail or give poor results, even with a well-trained legacy profile.

1. Press Win+I to open Settings.
2. Click Privacy & security in the left-hand sidebar.
3. Under Windows permissions, select Speech.
4. Turn the switch for Online speech recognition to On.
5. Go back to Privacy & security and select Voice activation.
6. Make sure apps are allowed to use voice activation. If you want this to work when the PC is locked, also allow it for the locked state.

Online speech recognition means your voice data goes to Microsoft’s cloud service. If that is a concern, leave the switch off and use Voice access, which recognizes speech on your PC, or the legacy Speech Recognition tools. To undo this change, return to the page and turn the switch off.

Personalizing Inking and Typing Dictionary

Windows can keep a personal dictionary of names, jargon and other words you use often. Turning on personalization lets that dictionary feed the speech-to-text engine, so unusual terms are less likely to be turned into the wrong word.

1. Open Settings and go to Privacy & security.
2. Scroll down to Windows permissions and click Inking & typing personalization.
3. Turn the switch for Personalization dictionary to On. Windows can then learn from how you type and speak and add custom words to your personal dictionary. Speech-to-text then transcribes your unique vocabulary more accurately.

On some Windows 11 builds this switch is labelled slightly differently, for example as a custom inking and typing dictionary. On some newer builds the page may be missing or reduced. If so, you will not be able to follow this step exactly on that PC.

To stop Windows learning from you, return to the page and turn the switch off. For a fixed list of special words, Voice access and its custom vocabulary are another option.

What to do next

Test your setup in a real application. Open a modern app such as Notepad or Word, click into the text area, and press Win+H to start voice typing. Dictate a few sentences that include the names and technical terms you use most. Then check how many words come out wrong.

Remember that Win+H dictation does not use the legacy trained profile. Your Control Panel training will not show up in these results. What you are testing here is the microphone, the language choice and the privacy permissions.

The voice typing window has its own settings. Click the gear icon in the voice typing toolbar to reach them. There you can turn Auto punctuation on or off, so Windows adds periods and commas for you, and control the mic launcher, which shows a small microphone button when you click into a text box.

You can also try Voice access (Settings › Accessibility › Speech) to confirm the microphone and permissions work. It uses on-device recognition, so it is a good comparison if the cloud-based voice typing performs badly.

If you see delays or missing words while you speak, check your internet connection first. Cloud-based speech processing needs network access, and a slow or dropped connection causes exactly those symptoms. If the connection is fine, go back and re-run the microphone test. Then repeat the voice training in a quieter room if you rely on the legacy tools.