Published September 10, 2026 · By Sumbat.T
Dictation Software for Windows 10 That Still Gets Updates
The short answer
Windows 10 has two built-in dictation tools and neither one is the good new one. Voice Access, the offline tool most articles mean when they say "built into Windows", is documented by Microsoft as Windows 11 version 22H2 and later, and Windows 10 passed end of support in October 2025. What Windows 10 has is Win+H, which Microsoft says needs an internet connection, and the older Windows Speech Recognition, which does not.
- Best for dictating into any app on Windows 10: BlabbyAI. Runs on Windows 10 and 11 alike, types wherever your cursor is, and punctuates as it goes. $8.49/month, with 60 credits a week free and no card.
- Best already installed: Windows voice typing. Press Win+H in any text box. Free, nothing to download, and it has an automatic punctuation setting. Microsoft: "to use voice typing, you'll need to be connected to the internet".
- Best with no internet: Windows Speech Recognition. The older Ease of Access tool, and the one to reach for when Win+H refuses because there is no connection. Its setup trains the recogniser to your own voice, and its accuracy is well behind anything built on a modern speech model.
- Best free and unmetered: Handy. MIT-licensed, runs the model on your disk, Windows x64 and ARM installers. Its own FAQ says spoken punctuation is "not directly" supported.
- Best for specialist vocabulary: BlabbyAI custom modes. A mode is a free-form instruction, so drug names, case citations and part numbers are handled by describing them in plain English rather than training a vocabulary file. Dragon Professional is the heavyweight alternative, and Nuance has retired self-serve purchasing, so it has no published price.
Windows 10 and Windows 11 alike
Dictation that does not care which Windows you are on
BlabbyAI types into any text field from the same installer on Windows 10 and 11, and the Chrome extension covers the browser with no download at all. Punctuates as it goes, and custom modes let you define your own spoken commands. 60 credits a week free, no card.
Why Windows 10 dictation advice is usually wrong
There is a specific mistake that runs through almost every article about dictating on Windows 10, and it is worth naming before anything else, because it wastes an afternoon. Guides tell you that Windows has a built-in accessibility tool called Voice Access, that it does on-device recognition, that it needs no internet, and that you should turn it on before paying for anything. All of that is true. None of it applies to you.
Voice access is available in "Windows 11, version 22H2 and later", and "uses modern, on-device speech recognition to accurately recognize speech and works even without the internet".
Microsoft Support, Voice Access system requirements, read 10 September 2026
Microsoft documents Voice Access as a Windows 11 22H2 feature, and Windows 10 reached end of support in October 2025, so nothing new is being added to it. When a roundup tells you the best free dictation software for Windows is the one built into Windows, check the version number in the requirement before you go looking for it in your settings. On a Windows 10 machine you will not find it, and nothing is broken.
What makes this more than a pedantic correction is the second half of it. Microsoft has stated that "on Windows 11 22H2 and later, Windows Speech Recognition (WSR) will be replaced by voice access starting in September 2024". Windows 11 got a replacement. Windows 10 kept the original, because there was nothing to replace it with. That is the whole shape of dictation on Windows 10 in one sentence: you have the old tool and the online one, and the good offline one went to the other operating system.
What you actually have, side by side
Four things get called "Windows dictation" in conversation and they behave differently enough that mixing them up is the source of most of the confusion. Here is what is on a Windows 10 machine and what is not.
| Tool | On Windows 10? | Works offline | Punctuation | Where it types |
|---|---|---|---|---|
| Voice typing (Win+H) | Yes | No, needs the internet | Automatic punctuation setting | Any text box |
| Windows Speech Recognition | Yes, via Ease of Access | The offline built-in option | Pre-AI engine, weakest here | Text plus interface control |
| Voice Access | No, Windows 11 22H2+ | Yes, on-device | Automatic | Text plus interface control |
| Word Dictate | Yes, with Microsoft 365 | No | Automatic | Microsoft Word only |
The surprise in that table is that voice typing needs the internet. Voice typing feels like a local feature: it has a keyboard shortcut, it opens instantly, and there is no obvious moment where anything leaves your computer. But the recognition happens in Microsoft's cloud, and the requirement is stated without hedging.
"To use voice typing, you'll need to be connected to the internet, have a working microphone, and have your cursor in a text box."
Microsoft Support, use voice typing to talk instead of type on your PC, read 10 September 2026
This is also, quietly, the answer to the most-asked Windows 10 dictation question there is, which is why Win+H does nothing. It is usually not a driver and not a microphone. It is that the tool is a cloud service and something in the path to that cloud is switched off.
Six reasons Win+H does nothing, in the order to check them
Microsoft Learn carries a question titled "How to turn on speech to text: windows key +H does not do anything", which ranks on page one for this problem: a good sign that it is common rather than something wrong with your particular PC. These are the causes worth ruling out, cheapest first.
01
Online speech recognition is off
The most common cause by a distance. Voice typing looks local but is not: it is a Microsoft cloud service behind a keyboard shortcut, so with the setting off the toolbar either refuses to open or opens and never transcribes. Turn it on in Settings under speech, then try Win+H again in a plain text box such as Notepad.
02
There is no internet connection
The same root cause with a different symptom. Microsoft requires a connection for voice typing because recognition runs in Azure Speech services. If you are on a train or an air-gapped work machine, Win+H is simply not going to work, and Windows Speech Recognition is the built-in tool that will.
03
The microphone is blocked in privacy settings
This one is unpleasant because the hardware tests fine everywhere else. Windows keeps a separate per-app microphone permission, and when it is off for the platform apps, dictation fails silently rather than showing an error. Check microphone privacy before you touch a single driver.
04
You are in an elevated or unusual window
Text insertion into a window running as administrator, an installer dialog, or some terminal emulators behaves differently from a normal text box. If dictation works in Notepad and fails in one specific application, the application is the variable, not the microphone.
05
The language is not installed
Voice typing needs the speech pack for the language you are dictating, and installing a display language does not always bring one. Microsoft points to Settings, Time and language, Language and region for adding it.
06
You expected Voice Access to be there
Not a fault at all, and worth ruling out first. If a guide told you to open Voice Access from accessibility settings and you cannot find it, you have not broken anything: it does not exist on Windows 10.
The one-minute test
Worth noticing about that list: four of the six causes exist because voice typing is a Windows feature gated on a Windows setting and a Microsoft cloud service. A dictation app you install yourself does not inherit any of them. It brings its own connection to its own service, it does not read the online speech recognition toggle, and it does not care whether the speech pack for your language was installed with your display language. BlabbyAI is one of those, and it runs on Windows 10 and Windows 11 from the same installer. The one cause that follows you across every tool is the microphone privacy permission, because that one is Windows deciding whether anything at all may listen.
What end of support actually changed, and what it did not
Windows 10 reached end of support on 14 October 2025, and Microsoft's lifecycle page records version 22H2 as the final version, with all editions supported through that date. Nearly a year on, that raises a reasonable question about whether dictating on Windows 10 is a dead end.
The practical answer is narrower than it sounds. End of support is about security updates to the operating system. Nothing was switched off: Win+H still opens the voice typing toolbar, Windows Speech Recognition is still sitting in Ease of Access, and Word's Dictate button still works if you have a Microsoft 365 subscription. What changed is the direction of travel. New speech features are being built for Windows 11, Voice Access is where Microsoft's effort is going, and none of it is being backported.
What this means for choosing a tool
Dictation software that runs on Windows 10 today
The modern generation of dictation tools does not depend on Windows version in the way the built-in features do, because the speech model is theirs rather than the operating system's. That is the reason a Windows 10 machine can run better dictation than a Windows 11 machine using only what came in the box.
| Tool | Price | Windows 10 | Runs where | Worth knowing |
|---|---|---|---|---|
| BlabbyAI | $8.49/mo, free tier 60 credits weekly | Yes, Windows 10 and 11 | Cloud | Windows 10 and 11 on one build, plus a Chrome extension |
| Wispr Flow | $15/mo, $12 billed annually | Windows desktop app | Cloud | Free tier 2,000 words a week on desktop |
| Willow Voice | $15/mo, $12 billed annually | Windows desktop app | Cloud | Free tier 2,000 words a week |
| Superwhisper | $8.49/mo, $84.99/yr, $249.99 once | Windows desktop app | On your machine | You choose and download a model |
| Handy | Free, MIT licensed | Yes, x64 and ARM installers | On your machine | Its own FAQ rules out spoken punctuation |
| Dragon Professional | No published self-serve price | Windows desktop app | On your machine | Deepest custom vocabularies, heaviest setup |
Every price in that table was read from the vendor's own pricing page on 7 September 2026, and the date matters more than it sounds in this category. The figures that circulate second-hand are frequently the annual rate quoted as though it were the monthly one, which makes a $15 tool look like a $12 one. Dragon Professional is the case where you cannot check at all: Nuance no longer sells it self-serve, so the prices you find are reseller listings rather than a published rate. We keep a dated price list for every tool in this category for the same reason.
The choice that actually matters: cloud or your own machine
Ignore the feature lists for a moment. The decision that determines how the tool feels day to day is where the speech model runs, and it splits the market into two halves that are hard to compare directly.
On your own machine
Superwhisper, Handy, OpenWhispr. The model file sits on your disk, nothing is uploaded, and it keeps working on a plane. The cost is setup: you choose a model, download a few hundred megabytes, and own the troubleshooting when one does not suit your hardware. Handy's own README lists model crashes on certain Windows configurations as a known issue.
Best when the machine never leaves a network you control, or when the text is confidential enough that the question has to have a structural answer.
In the cloud
BlabbyAI, Wispr Flow, Willow, and Windows' own Win+H. Nothing to choose and nothing to download beyond the app itself, accuracy improves without you doing anything, and an old Windows 10 PC is not asked to run a speech model it does not have the hardware for. The trade is that it needs a connection.
Best on older hardware, which describes a lot of machines still running Windows 10, and when you would rather not maintain anything.
That second point deserves more weight than it usually gets in Windows 10 advice specifically. If your PC is still on Windows 10 because it did not meet Windows 11's hardware requirements, then it is by definition a machine Microsoft considers below the current bar, and asking it to run a multi-gigabyte speech model locally is a worse experience than sending a few seconds of audio to a server. Guides that recommend fully local tools as the obvious privacy-conscious choice rarely make that connection, and on this particular audience it matters more than anywhere else.
Where BlabbyAI fits on a Windows 10 machine
BlabbyAI supports Windows 10 and Windows 11 with the same application, so nothing about this is a compatibility mode or a cut-down build. You press a shortcut, speak, and the text appears wherever your cursor already is. That is the same shape as Win+H, which is deliberate: the shortcut is the interaction people already know.
Three differences are worth spelling out, because they are the reasons somebody moves off the built-in tool rather than marketing points.
- Punctuation arrives without being spoken. Win+H has an automatic punctuation setting, which Microsoft describes as letting voice typing "automatically insert punctuation based on what you're saying". BlabbyAI punctuates as a matter of course, and its custom modes go further: a mode is a free-form instruction, so you can define spoken commands of your own such as "new paragraph" and have them act rather than appear as words.
- It types into every application, not a list of them. Word's Dictate button lives in Word. Google Docs voice typing lives in Google Docs in Chrome. BlabbyAI integrates with whatever has a text field: Outlook, Teams, Slack, Notion, Todoist, a code editor, an EHR. It integrates with Epic the same way it integrates with everything else: it types into any Epic text field, with no plugin needed.
- There is a browser option with no download at all. A good share of what anyone types on a Windows 10 PC happens in a browser tab, and the Chrome extension is a product in its own right rather than a lesser version of the desktop app: Gmail, Google Docs, Notion, Jira, a CRM, any text field in the tab. It needs no installer and no admin rights, which also makes it the answer on a locked-down work machine.
What it costs
No installer? No problem
Dictate in the browser on any Windows 10 PC
The BlabbyAI Chrome extension needs no download and no admin rights, and types into Gmail, Google Docs, Notion and every other text field in the tab. Free to start.
How to decide, in about two minutes
The honest version of this is short, because most of the branches collapse quickly once you know Voice Access is off the table.
- You dictate a few times a week and have internet. Win+H is free and already installed, and for a paragraph of email it is fine. If you would rather not speak your punctuation or watch the toolbar drop out when you change window, BlabbyAI's free tier covers that much dictation indefinitely at 60 credits a week.
- You dictate often, across several applications. A system-wide third-party tool pays for itself in not fighting punctuation. BlabbyAI at $8.49 is the cheapest of the maintained cloud options; Wispr Flow and Willow are $15.
- The machine has no reliable internet. Windows Speech Recognition is your free built-in answer, and Superwhisper or Handy are the modern ones. Accept that you are choosing and maintaining a model.
- You cannot install software on the machine. A Chrome extension is the only category that solves this, and it covers anything you do in a browser.
- You dictate a lot of specialist vocabulary. Drug names, case citations, part numbers and surnames are where generic dictation falls down. A BlabbyAI custom mode handles this as an instruction in plain English ("always spell our product names this way"), which is a different mechanism from Dragon Professional's trained vocabulary files and a much shorter setup.
Frequently asked questions
Does Windows 10 have built-in dictation software?
Yes, two separate ones, and they are not the same tool. The first is voice typing, which you open with Windows logo key + H in any text box. It is the modern one, and Microsoft states plainly that "to use voice typing, you'll need to be connected to the internet, have a working microphone, and have your cursor in a text box." The second is Windows Speech Recognition, the older engine reached through Ease of Access, which runs on your machine and can also drive menus and windows by voice. Neither is Voice Access, which is the good offline one and is Windows 11 only.
Can I get Voice Access on Windows 10?
No. Microsoft documents Voice Access as available in "Windows 11, version 22H2 and later", and it has never shipped for Windows 10. This is the single most common misunderstanding about dictation on Windows 10, because most roundups describe Voice Access as "the built-in Windows dictation tool" without noting the version requirement. Voice Access uses on-device speech recognition and works without the internet; Windows 10 users get the online-only Win+H instead. If you want on-device style convenience without changing operating systems, that gap is what third-party dictation software fills.
Does Windows 10 dictation work offline?
Voice typing (Win+H) does not. Microsoft's own requirement is an internet connection, because the recognition runs in Azure Speech services rather than on your PC. Windows Speech Recognition, the older Ease of Access tool, is the one to reach for when there is no connection: its setup walks you through training the recogniser to your own voice, which is not something a cloud service asks you to do. The trade is that it was built long before modern speech models, so accuracy and punctuation are well behind what Win+H or any current third-party tool gives you.
Is Windows 10 dictation still supported now that Windows 10 has ended?
Windows 10 reached end of support on 14 October 2025, and version 22H2 was the final version. That is about security updates for the operating system rather than a switch being thrown on features: Win+H still opens, and Windows Speech Recognition is still in Ease of Access. What it does mean is that nothing new is coming. Voice Access, auto-punctuation improvements and the newer language additions are landing on Windows 11 and will not be backported. Anything that gets better on Windows 10 from here is going to be third-party software, which is why the question is worth asking at all.
What is the best free dictation software for Windows 10?
For occasional use, the one already installed: Win+H costs nothing, needs no download, and handles a paragraph of email perfectly well. For regular use, the honest free options are metered rather than unlimited. BlabbyAI gives 60 credits a week, roughly 2,000 words, with no card required, and runs on Windows 10 and Windows 11. Wispr Flow gives 2,000 words a week on desktop. Handy and OpenWhispr are free and open source with no cap at all, which is genuinely generous, but you pick and download a speech model yourself and own the troubleshooting when a particular model does not agree with your hardware.
Why does Win+H do nothing on my Windows 10 PC?
Almost always one of three things, and it is common enough that a Microsoft Learn question titled "How to turn on speech to text: windows key +H does not do anything" ranks on page one for the problem. First, online speech recognition is switched off: it lives in Settings, and voice typing will not start without it because the recognition happens in the cloud. Second, there is no internet connection, which is the same root cause with a different symptom. Third, the microphone is blocked at the privacy level rather than the hardware level, so the device works everywhere else and silently fails here. Worth knowing before you spend an evening on drivers: if the shortcut opens the toolbar but no words appear, it is nearly always the second or third one.
Does dictation software work in every app on Windows 10?
It depends on how the tool is built. Win+H and BlabbyAI both type into whatever text field currently has your cursor, so Word, Outlook, Chrome, Slack, Notion, a code editor and an EHR text box are all just text fields to them. Tools built as an add-in are narrower: Word's own Dictate button works in Word and nowhere else, and Google Docs voice typing works in Google Docs in Chrome and nowhere else. If you dictate across several applications in a day, a system-wide tool saves you learning three different shortcuts.
Do I need to upgrade to Windows 11 to get good dictation?
No, and this is worth saying plainly because a lot of advice quietly assumes it. Modern dictation quality now comes from the speech model, not from the operating system, and third-party tools run that model either in the cloud or on your own machine regardless of which Windows you are on. BlabbyAI supports Windows 10 and Windows 11 equally. The one thing Windows 11 gives you that Windows 10 cannot have is Voice Access, and Voice Access is as much a voice-control tool for driving the interface as it is a dictation tool.
How much does dictation software for Windows 10 cost?
Between nothing and about $17 a month, and all of the following were read from each vendor's own pricing page on 7 September 2026. BlabbyAI is $8.49 per month for unlimited use, the same figure whether you pay monthly or yearly, with a free tier of 60 credits a week. Superwhisper is also $8.49 monthly, $84.99 for a year, or $249.99 once for a lifetime licence. Wispr Flow and Willow Voice are both $15 monthly and $12 billed annually. Otter.ai is $16.99 per user per month, though it records meetings rather than typing where your cursor is. Dragon Professional no longer has a published self-serve price at all.
Is Windows Speech Recognition being removed?
On Windows 11 it already has been, in effect: Microsoft states that "on Windows 11 22H2 and later, Windows Speech Recognition (WSR) will be replaced by voice access starting in September 2024." On Windows 10 there is no replacement to be moved to, so WSR is still there in Ease of Access. That is the odd position Windows 10 ends up in: it keeps the older tool precisely because it never received the newer one.
Dictation that does not depend on your Windows version
BlabbyAI runs the same on Windows 10 and Windows 11, types into whatever text field your cursor is in, and punctuates without being asked. Custom modes let you define your own spoken commands, such as saying 'new paragraph' to get a real paragraph break. 60 credits a week free, no card.
Sources
- Microsoft Support, "Use voice typing to talk instead of type on your PC". The internet requirement and the automatic punctuation setting. Read 10 September 2026.
- Microsoft Support, "Use voice access to control your PC". The Windows 11 22H2 requirement, the on-device recognition, and the replacement of Windows Speech Recognition. Read 10 September 2026.
- Microsoft Learn, Windows 10 Home and Pro lifecycle. End of support on 14 October 2025 and 22H2 as the final version. Read 10 September 2026.
- Vendor pricing pages for BlabbyAI, Superwhisper, Wispr Flow, Willow Voice and Otter.ai, read 7 September 2026.
- BlabbyAI, Windows speech to text and dictation software.