How to Change Default Voice in Windows 11 and 10 Speech Recognition

To change the default text-to-speech voice in Windows 11 or Windows 10, press Win + I, open Time & language → Speech, and pick a voice under Text-to-speech. But here is the part Microsoft never says out loud: the beautiful natural voices Windows advertises cannot be your default voice. They exist only on Windows 11 22H2 or newer, install through Narrator, and are locked to Narrator — they will never appear in that drop-down, and no language pack changes it.

If you opened this page because David, Zira, or Mark is grinding through your documents like a robot from 1998, you are not doing anything wrong — those really are the only voices Windows offers in that drop-down until you add more, and even then the additions are cousins of the same classic engine. This guide walks through changing the default the right way on both Windows versions, adding every voice that can legally appear in that list, and — just as important — being straight with you about where Microsoft's genuinely human-sounding voices live and how to actually use them.

Why the Voice Drop-Down Never Shows the Good Voices

Windows quietly runs two separate voice worlds, and almost every frustrated search about text-to-speech comes from not knowing the border between them. The first world is the classic Speech API (SAPI) family: Microsoft David, Zira, Mark, and the voices that arrive with language packs. These are the voices the Time & language → Speech page controls, the ones Control Panel's old Speech Properties dialog sets, and the ones ordinary desktop programs ask for when they speak.

The second world is the natural voices — the warm, human-sounding ones Microsoft shows off. They run on a completely different neural engine, they exist only on Windows 11 version 22H2 and newer, and they are installed from Settings → Accessibility → Narrator, not from language packs. Most important of all: they are wired to Narrator alone. They never register in the classic voice list, so the Speech page drop-down, Word's Read Aloud picker on perpetual Office versions, and nearly every third-party reading app simply cannot see them.

My friend Ethan fixes and sets up office PCs for a living, and he has watched more people fight this wall than any other speech problem. His summary is the one worth remembering: the Settings page changes the voice for the old world, Narrator's own settings change it for the new world, and no amount of downloading language packs moves a voice from one world to the other.

The good news hiding inside that bad news: you do not need the system default to hear natural speech. Microsoft Edge's Read aloud streams its own excellent online voices on both Windows versions, Microsoft 365 apps carry their own large online voice sets, and Narrator itself becomes a genuinely pleasant document reader once a natural voice is installed. The rest of this guide covers each door, in order of how likely you are to need it.

How to Change the Default Voice in Windows 11

This is the change most people came for: swapping which classic voice speaks by default, and how fast it talks. It is a per-user preference, needs no administrator rights, takes effect immediately, and cannot harm your files or your Windows license — you can switch back at any moment.

1. Press Win + I to open Settings.
2. Click Time & language in the left-hand pane.
3. Click Speech.
4. Under Text-to-speech, open the Voice drop-down and choose a voice — for US English installs that usually means Microsoft David, Microsoft Zira, or Microsoft Mark.
5. Click Preview voice to hear a sample sentence in your selection.
6. Drag the Voice speed slider until the pace feels right, and preview again.

There is no Save button; the selection is stored the moment you make it. If the drop-down offers only one or two voices, that is normal for a fresh install — the section on adding more voices below fixes it in a few minutes.

How to Change the Default Voice in Windows 10

Windows 10 keeps the same layout with slightly different labels, and everything on this page still works exactly as it always did — text-to-speech runs from local voice files and does not depend on updates or an internet connection.

1. Press Win + I to open Settings.
2. Select Time & Language.
3. Click Speech in the left-hand column.
4. Find the Text to speech section on the right.
5. Open the Voice drop-down and choose your preferred voice.
6. Click Preview voice to confirm it sounds the way you expect, and adjust the speed slider to taste.

If parts of this page appear grayed out on a work machine, a domain policy set by your IT department is usually the reason rather than anything wrong with your account — the same steps performed by an administrator account, or a polite ticket to IT, resolves it. On a home PC, a standard account can make this change freely.

One thing Windows 10 users should know before chasing voices: the natural voices covered later in this guide do not exist on Windows 10 at all — not through language packs, not through the Store, not through Narrator. The honest best option for natural-sounding reading on Windows 10 is Microsoft Edge's Read aloud, covered below, which works beautifully on Windows 10 because Edge streams its own voices.

How to Add More Voices to the Drop-Down

The stock list is short on purpose — Windows only installs voices for your setup language. Two supported ways to grow it, both free:

The direct way: on the same Speech settings page, look for the Manage voices section and click Add voices. Search or scroll to a language or regional variant — English (United Kingdom), English (India), and so on — select it, and click Add. Windows downloads the voice data, and after the download finishes the new voices join the same Voice drop-down you just used. Each variant typically brings one or two voices, such as Hazel for UK English or Ravi and Heera for Indian English.

The language-pack way: go to Settings → Time & Language → Language & region, click Add a language, and when choosing optional features make sure Text-to-speech is checked. This installs the language's voices along with keyboard and display resources. It is the right route when you also want to type or display that language; if you only want the voice, the Add voices button is lighter.

Be clear about what you are getting: these additions are more classic voices — useful for other accents and languages, and some people find one they much prefer — but they are built on the same older engine as David and Zira. Adding language packs does not, on any Windows version, install or unlock the natural neural voices. If an article told you otherwise, it was wrong, and this page used to be less clear about that than it should have been.

Which Voices Work Where: The Map Microsoft Never Drew

Every disappointment with Windows text-to-speech traces back to this table. Bookmark it before you spend an evening installing things that were never going to appear where you hoped.

The Classic Control Panel Speech Properties Route

The old Speech Properties dialog from the Windows 7 era never went away — Microsoft just stopped linking to it. It still exists on both Windows 11 and Windows 10, and it matters because some older desktop programs read their voice from this classic setting rather than the modern Settings page.

To open it, press Win + R, type sapi.cpl, and press Enter. On the Text to Speech tab, choose a default under Voice selection, type any sentence into the box and click Preview Voice to hear it, set the Voice speed, and click OK.

There is a second, even more obscure copy for 32-bit programs on 64-bit Windows: run %systemroot%\syswow64\speech\speechux\sapi.cpl to open the 32-bit Speech Properties. If an old 32-bit reading program stubbornly uses the wrong voice no matter what you set elsewhere, this dialog is usually the one it is listening to.

A practical habit Ethan recommends when a program ignores your chosen voice: set the same default in all three places — the Settings page, the 64-bit Speech Properties, and the 32-bit one — and the mystery almost always disappears, because every app family is now reading the same answer.

Installing the Natural Voices in Narrator (Windows 11 22H2 and Newer)

If what you actually want is a human-sounding voice reading your screen, this is where Microsoft put it. The natural voices download once and then run entirely on your device, so they keep working offline — and inside Narrator they are genuinely good, close to the audiobook standard people expect in 2026.

1. Press Win + I and go to Accessibility → Narrator.
2. Under Narrator's voice, click Add natural voices (the button appears on Windows 11 22H2 and newer).
3. Browse the list, click Preview on any voice to hear a sample before downloading.
4. Select the voice you want and click Install; the download runs in the background.
5. Once installed, pick the natural voice from the Choose a voice drop-down in Narrator's settings, and adjust Narrator's own speed and pitch controls there.
6. Start or stop Narrator any time with Win + Ctrl + Enter.

Two expectations to set honestly. First, the selection started in US English and has grown to cover many more languages over time, but not every language has a natural voice yet — the Add natural voices list on your own PC is the truthful catalog for your region. Second, and this is the heart of this whole article: installing a natural voice changes Narrator and nothing else. It will not appear in the Speech settings drop-down, in Speech Properties, or in Word's voice picker. That is a design decision by Microsoft, confirmed repeatedly in their own support forums, not a bug you can fix by reinstalling.

Word and Edge Read Aloud: Where the Good Voices Actually Are

Proofreading by ear is one of the best editing tricks there is — a flat robotic voice makes a 40-page contract feel endless, while a natural one catches your missing words for you. Here is where each app really gets its voice, because none of them blindly follow the system default.

Microsoft Word and Outlook: start Read Aloud from the Review tab (or Ctrl + Alt + Space in Word), then click the gear icon on the small playback toolbar. The Voice Selection drop-down there is the setting that counts, along with its own reading-speed slider. On Word 2019 and 2021 that list shows the classic system voices — so the Windows Speech settings decide what is available, and adding voices there does grow Word's list. On Microsoft 365, Word offers its own far larger set of streamed online voices, which is where Word finally sounds modern.

Microsoft Edge: open any web page or PDF, press Ctrl + Shift + U or click Read aloud, then open Voice options on the toolbar. Edge streams Microsoft's online natural voices — a long list across many languages — and this works identically on Windows 10 and Windows 11, because the voices come from Edge, not from Windows. This is the single best free answer for Windows 10 users who want natural speech today; the trade-off is that streamed voices need an internet connection.

When Jake first heard the difference, his question was the obvious one: why does his six-year-old laptop sound like a podcast host in Edge and like a fax machine in everything else? Same machine, different worlds — Edge brings its own voices with it. Drop a local PDF into Edge and you have effectively given every document on your PC a natural voice, no settings changed at all.

The Registry Trick That Reveals Hidden Voices

Here is a genuinely useful secret for the classic world: Windows often has more voices installed than it shows you. Voices that arrive with language packs register under a modern branch, HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech_OneCore\Voices\Tokens, but many older programs — and the classic Speech Properties dialog — only read the legacy branch, HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens. Copying a voice's token key from the OneCore branch to the legacy branch (and to HKEY_LOCAL_MACHINE\SOFTWARE\WOW6432Node\Microsoft\Speech\Voices\Tokens for 32-bit programs) makes that voice appear in the classic lists.

Before you touch the registry

This trick has circulated since the Windows 10 days and, as of this rewrite, the same token-copy approach is still what enthusiast tools and forum guides use on current Windows 11 builds — but Microsoft has never officially supported it, and a future update could change how these branches are read. So take the two-minute safety net: in Registry Editor, right-click the Speech and Speech_OneCore keys and choose Export to save .reg backup files first. Undoing the trick is then just deleting the token keys you added, or double-clicking your backups.

The practical method: open regedit as administrator, navigate to the OneCore Tokens key, right-click the voice you want (its name looks like MSTTS_V110_enGB_HazelM), choose Export, open the exported .reg file in Notepad, replace every Speech_OneCore path with Speech, save, and double-click it to merge. Repeat with the WOW6432Node path for 32-bit apps. The voice then shows up in the Voice drop-down and in Speech Properties like any other.

Know the hard limit before you start: this only reveals hidden classic-engine voices. It cannot move a natural voice into the SAPI list — the natural voices are a different engine, not a hidden token. An open-source project called NaturalVoiceSAPIAdapter tries to bridge that gap for adventurous users, but it is unofficial, unsupported by Microsoft, and can break with updates; we mention it for completeness, not as a recommendation.

The Voice I Installed Doesn't Show Up: Symptoms and Fixes

Nearly every "it didn't work" report about Windows voices lands in one of the rows below. Find your symptom first; most of these are five-minute fixes, and none of them mean your PC is damaged.

If you have walked the whole table and the speech stack still misbehaves — empty lists that survive a reinstall of the language, previews that error out everywhere — the honest last resort is an in-place repair install of Windows (running setup from within Windows and choosing to keep files and apps). It rebuilds system components, including speech, without touching your documents or programs. It takes an hour, not an evening, and it beats chasing corrupted components one registry key at a time.

Still on Windows 10? What Keeps Working After End of Support

Windows 10 reached its official end of support on October 14, 2025, and an enormous number of PCs are still running it every day — if yours is one of them, nothing about this guide stops working. Text-to-speech runs from voice files already on your disk, so changing voices, previewing them, adding language voices, and the registry trick all keep functioning with or without updates.

If you are keeping a Windows 10 machine online, it is worth knowing that Microsoft's consumer Extended Security Updates program now runs through October 12, 2027 — Microsoft extended it by a year in June 2026 — and enrollment is free via settings backup sync, or available for Microsoft Rewards points or a small one-time fee, covering up to ten devices per license. That has nothing to do with voices directly, but it decides whether the PC you are configuring stays safe to use.

The voice-specific reality for Windows 10 is the one this article has repeated on purpose, because so many guides get it wrong: the natural voices will never come to Windows 10. Microsoft built them into Windows 11 22H2's Narrator and stopped there. Your natural-speech options on Windows 10 are Edge's streamed Read aloud voices, the online voices inside Microsoft 365 apps if you subscribe, or upgrading the machine to Windows 11 where hardware allows.

None of that makes the classic voices useless. For alarm-style notifications, quick selection reading, and older accessibility software, David and Zira remain perfectly serviceable — and after the Add voices step you may find a regional voice that suits your ear far better than the defaults ever did.

Frequently Asked Questions

How do I change the default voice in Windows 11?

Open Settings with Windows + I, go to Time & language, click Speech, and pick your preferred voice from the Voice drop-down under Text-to-speech. Use the Preview voice button to hear it before you commit, and drag the Voice speed slider to adjust the pace. The change applies immediately with no restart.

Does changing the Windows voice change Microsoft Word Read Aloud?

Not directly. Word's Read Aloud has its own voice picker: start Read Aloud, click the gear icon on the small toolbar, and choose from the Voice Selection drop-down. The Windows Speech settings decide which classic voices are available to Word 2019 and 2021, but Word does not simply follow the system default, and Microsoft 365 versions of Word offer their own much larger set of online voices.

How do I open speech settings on Windows 10?

Press Windows + I, select Time & Language, then click Speech in the left-hand column. The Text to speech section on the right holds the Voice drop-down, the Preview voice button, and the speed slider.

Can I get the natural voices on Windows 10?

No. Microsoft's natural voices require Windows 11 version 22H2 or newer and are installed through Settings > Accessibility > Narrator, not through language packs. On Windows 10 your best route to natural-sounding reading is Microsoft Edge's Read aloud feature, which streams its own high-quality online voices and works on Windows 10 without any system changes.

Why don't the natural voices show up in the Voice drop-down?

Because Microsoft built them for Narrator only. The natural voices never register in the classic Speech voice list, so the Time & language > Speech drop-down, Control Panel Speech Properties, Word's Read Aloud picker, and most third-party apps cannot see them. This is by design, not a fault on your PC, and no language pack or setting changes it.

Does Windows Narrator use the same voice settings?

No. Narrator has its own voice, speed, and pitch controls under Settings > Accessibility > Narrator, and that is also the only place the natural voices can be selected. Changing the voice on the Time & language > Speech page does not change the voice Narrator reads with.

Do text-to-speech voices need an internet connection?

The classic voices such as David, Zira, and Mark work entirely offline. Narrator's natural voices need internet once for the download, then run fully on-device and keep working offline. Edge's Read aloud voices and the online voices in Microsoft 365 Word are streamed, so those do need an active connection.

How do I preview a voice before applying it?

On the Speech settings page, click the Preview voice button just below the Voice drop-down and Windows speaks a sample sentence in the currently selected voice. In Control Panel's Speech Properties you can type your own sentence into the text box and click Preview Voice to hear exactly how it handles your words.

Why did my voice settings revert after a Windows update?

Major feature updates occasionally reset speech preferences or default audio device associations. Nothing is lost: reopen Settings > Time & language > Speech and reselect your voice, and it sticks again. If you used the registry method to reveal extra voices, check that those keys survived the update too.

What happened to the Control Panel speech options?

They still exist, just hidden. Press Windows + R, type sapi.cpl, and press Enter to open the classic Speech Properties dialog on both Windows 11 and Windows 10. It sets the same default that older desktop programs read, so it remains useful when a legacy app ignores the modern Settings page.

Does changing the text-to-speech voice affect speech recognition?

No. Text-to-speech voices control how your computer speaks to you, while speech recognition profiles control how it understands your microphone input. The two are separate systems, and changing one never degrades the other.

Is the registry trick to unlock extra voices safe?

It has a long track record and only copies voice token keys from one registry branch to another, but any registry edit deserves a safety net. Export the Speech and Speech_OneCore branches to .reg files first so you can restore them, and undo the change by deleting only the token keys you added. It reveals hidden classic voices only; it cannot unlock the Narrator natural voices.

How do I remove voice packs to free up space?

Go to Settings > Time & Language > Language & region, click the three-dot menu next to the language you added, choose Language options, and remove the Text-to-speech feature or the whole language pack. Narrator natural voices are removed separately from Settings > Accessibility > Narrator.

Do I need administrator rights to change the voice?

Not for the basic change: picking a voice on the Speech settings page is a per-user preference and works on a standard account. You will need administrator rights to install new language packs on some managed PCs and for the registry method, since those voice token keys live under HKEY_LOCAL_MACHINE.

Can each user account have a different voice on a shared PC?

Yes. The default voice selection is stored per user profile, so family members or coworkers sharing one machine can each pick their own voice and speed without affecting anyone else. Installed voice packs themselves are system-wide, so a voice one person installs becomes available to everyone.

Are the extra voices free, or will Microsoft charge me?

Every voice discussed here is free: the classic voices, the language-pack voices, the Narrator natural voices, and Edge's Read aloud voices all cost nothing. The only voices tied to a paid product are the online voices inside Microsoft 365 apps, which come with the subscription rather than as a separate charge.

Why does Edge Read aloud sound so much better than my system voice?

Edge does not use your system voice at all. Its Read aloud feature streams Microsoft's own online natural voices, which are far ahead of the classic offline voices. That is also why Edge sounds great even on a Windows 10 PC that cannot install natural voices.

Is there any way to use a natural voice as the system default?

Not with anything built into Windows. A third-party open-source project called NaturalVoiceSAPIAdapter attempts to bridge natural voices into the classic voice list, but it is unofficial, unsupported by Microsoft, and can break with Windows updates, so treat it as an experiment rather than a fix. For everyday reading, using natural voices where they already live, in Narrator, Edge, and Microsoft 365 apps, is the reliable path.

Revision note. Originally published November 2016. Rewritten August 2026 for Windows 11 and Windows 10. Back then, every voice lived in one Control Panel dialog; today Windows runs two separate voice systems, and an earlier update of this page overstated what language packs can do — they cannot unlock the natural voices, which work only inside Narrator on Windows 11 22H2 and newer, and this version sets that straight. If you arrived here tired of a robotic voice reading your work, we hope you leave with one you actually enjoy listening to — and if something on your machine still refuses to cooperate, the failure table above was written for exactly that moment. We are right here with you.