Windows Speech Recognition not working

Half the time it's deprecated, not broken

The old app is Windows Speech Recognition, WSR for short. It's the legacy desktop feature you start with Windows logo key + Ctrl + S, with the blue control bar across the top of the screen and a setup wizard that asks you to read a paragraph aloud so it can learn your voice. That's the thing most "WSR not working" searches are about. And the most common reason it doesn't work is that it's not there anymore.

Microsoft's own support page is blunt about it: Voice access replaced Windows Speech Recognition for Windows 11 version 22H2 and later, back in September 2024. So if you're on a current Windows 11 and the WSR app won't open, that's not a fault you can troubleshoot. It's a feature that was retired. It still ships on Windows 10 and on older Windows 11 builds, which is why the experience splits so hard depending on your version.

Then there's the confusion tax. Voice typing, the bar you summon with Windows logo key + H, is a completely separate feature. It's the modern, in-any-text-box dictation most people actually want, and it's alive and well. If you press Win+H and it works, you may never have needed WSR in the first place. So step one of fixing WSR is figuring out whether you want WSR at all.

Fix 1: check whether your version still has it

Press Windows logo key + Ctrl + S. On Windows 10, or an older Windows 11, the blue WSR control bar appears and you can run the setup wizard. On Windows 11 version 22H2 and later, you'll find the old app is gone and Windows steers you toward Voice access instead. If that's what you see, WSR isn't broken on your machine, it has been replaced, and no amount of poking at settings will bring it back.

To confirm what your version actually supports, Microsoft's own Speech Recognition page lays out the split: separate instructions for Windows 10 and Windows 11, and the note that Voice access is the successor on current Windows 11. Check your build under Settings, System, About if you're unsure which side of the line you're on.

If you're on Windows 10 and the app does open, the rest of these fixes apply to you. If you're on a newer Windows 11 and it's gone, skip ahead to the modern replacements section, because that's the real answer for your version.

Fix 2: your mic isn't set up, or permission is off

If WSR does open but hears nothing, the problem is almost always the microphone, not the speech engine. Open Settings, System, Sound, Input, pick the device you actually speak into, and watch the input-level bar. Talk. If the bar doesn't move, Windows is pointed at the wrong mic, or the one it's using is muted. This catches more "speech recognition not working" cases than anything else.

The second half is permission. Open Settings, Privacy and security, Microphone, and confirm microphone access is on for the device, and that desktop apps are allowed to use it. Speech features fall under that catch-all toggle, and when it's off, WSR records silence and gives up. Both toggles on, right device selected, level bar bouncing: that's the trio that has to be true before anything else matters.

One more boring but real cause: a USB mic or headset that disconnected and reconnected on a different port. Windows sometimes treats it as a new device and quietly switches the default back to the laptop's built-in mic. If your dictation worked yesterday and not today, and you moved a cable, start there.

Fix 3: language or region mismatch, and re-running setup

WSR is fussy about language. If your Windows display language, your keyboard layout, and the speech language don't line up, the recognizer can refuse to start or sit there understanding nothing. Open Settings, Time and language, Language and region, and make sure the language you actually speak is installed and set as the Windows display language, with its speech pack present. A mismatch here is a quiet, common reason WSR "works" but transcribes gibberish.

If the app opens but recognition is poor or stalled, re-run the setup wizard. From the WSR control bar, the configuration and "Train your computer to better understand you" options walk you back through microphone setup and a short voice-training pass. On a fresh Windows profile, or after a big feature update, that re-training is sometimes what it takes to get the legacy engine cooperating again.

I'll be honest about the ceiling here, though. Even when you get WSR fully working, it's a recognizer from a different era. The training wizard exists because the model needs your help; modern engines don't ask. So if you find yourself reading paragraphs aloud to calibrate an app in 2026, that's a sign worth listening to. There's a better road, and it's the next two sections.

The modern replacements: Win+H and Voice access

Here's the good news buried under the bad. The feature most people wanted from WSR has already moved somewhere better. Press Windows logo key + H in any text box and the Voice typing bar appears, transcribes what you say, and drops it straight into whatever you're writing. No training wizard, no blue control bar, no setup ritual. For everyday dictation, this is the answer, and it's the one Microsoft is actually maintaining.

For full hands-free control, where you also want to move the cursor, click buttons, and run commands by voice, Voice access is the WSR successor on Windows 11. You turn it on under Settings, Accessibility, Speech. It does the command-and-control job the old app did, just built for current Windows. If you came to WSR for accessibility rather than plain dictation, Voice access is where that capability lives now.

So the version question from earlier resolves cleanly. On a current Windows 11, don't fight to revive WSR. Use Win+H for dictation, Voice access for hands-free control, and treat the old app as the museum piece it has become.

If you are choosing between free Win+H Voice typing and a paid heavyweight, our Dragon vs Windows Voice Typing comparison weighs accuracy, cost, and setup so you can skip the trial-and-error.

A dictation app that doesn't depend on WSR at all

If you'd rather not be at the mercy of which Windows feature is alive this year, Whisper by Remskill takes the dependency away entirely. It's a desktop app that brings its own speech engine, so it doesn't matter whether WSR is present, deprecated, or carried off in the night. Press a hotkey, speak, and the transcript pastes wherever your cursor is, in any app.

On Windows the default hotkey is Ctrl+Space; on a Mac you hold Command+Option together and release either key to stop. Because the transcription runs on your own machine, pure-Rust, no Python sidecar, no server in the loop, it works offline and your words never leave your laptop. There's no online speech toggle to forget and no cloud endpoint that can quietly go away.

That's the structural difference from a built-in Windows feature. WSR could be removed in an update because it was part of Windows. A maintained app's whole job is to keep working when the OS reshuffles, which, as the last two Windows 11 releases have shown, it does.

What the full Whisper app looks like

The hotkey is the part you'll use every day, but there's a settings surface behind it. You pick your transcription engine: Whisper models, whose multilingual variants cover 99 languages and whose English-only builds cover exactly one, or NVIDIA's Parakeet, about 600 MB and 5 to 10 times faster than Whisper on CPU, covering English plus 24 European languages. You set the hotkey, manage history, save presets. None of it is required to dictate one sentence.

An optional AI cleanup step runs locally through Ollama, free, on your own machine. It tidies punctuation and trims the filler before pasting, so "um, so the report's, the report's done I think" comes out as "The report's done." You can turn it off and paste the raw transcript. Both are fine.

When the built-in is actually enough

Here's where I tell you not to install anything. If all you do is dictate the occasional short note, and you're on a version of Windows where Win+H works, just use Win+H. It's free, built in, and good enough for a quick email or a search box. There's no reason to add an app for something the OS already does for the once-a-week case. I'd rather you knew that than felt sold to.

Where a dedicated app pulls ahead is the rest of it: working the same way in every app, dictating offline with no internet at all, the punctuation cleanup, and not being one Windows update away from your dictation feature disappearing. The longer and more often you write by voice, the more those add up. For a single short note, the built-in is the right call. For a daily writing habit, the app earns its place.

What Whisper costs

The local dictation pipeline, transcription and the AI cleanup over Ollama, is free for any signed-in user, with no card at signup. So getting your voice onto the page with Whisper costs nothing. Whisper Pro adds the cloud features (OpenAI transcription, cloud AI enhancement, voice web search) on a separate trial. The exact numbers live on the pricing page rather than here, because prices move and a blog post is a bad place to keep them current.

Windows Speech Recognition isn't really a thing you fix anymore. It's a thing you replace. On Windows 10 it still runs, and the mic-and-permission checks above will get it going. On current Windows 11 it's gone, replaced by Voice access, and chasing it is chasing a ghost. Either way the durable move is the same: dictate with Win+H, or with an app that brings its own engine and keeps working no matter what Microsoft does to the OS this year.