Dictation software that works with screen readers on Windows

Most coverage of screen readers and dictation treats compatibility as a yes or no. It is really two questions: can you start dictation while your screen reader is running, and does the text it produces land where you are typing? We tested the popular Windows dictation apps against both, on NVDA, JAWS, and Narrator, and the answers split the field.

What we tested

NVDA, JAWS, and Windows Narrator, on Windows 11

The two axes

Does the hotkey fire, and does the text land, with a screen reader on

Best overall

Whisperstream for private one-time dictation; Voice Access for a free built-in option

Screen-reader compatibility comes down to two separate questions, and an app can pass one and fail the other. The first is the activation hotkey. A screen reader reserves certain keys as its own modifier (Insert and CapsLock on NVDA and JAWS, and the same pair on Narrator), so a dictation app whose push-to-talk shortcut leans on one of those can find its hotkey swallowed before recording ever starts. That is what we saw with Aqua Voice and Narrator: the dictation itself was fine, but the default shortcut would not fire.

The second question is text injection: once you have spoken, does the text actually land in the focused field? Here the method matters. A paste bound to Shift plus Insert is handy because it works in terminals, but Shift plus Insert is exactly the chord a screen reader claims for its own commands, so the paste can do nothing. Apps that paste with Ctrl plus V, insert text directly at the cursor, or use the Windows dictation API sidestep that collision, which is why several cloud and local apps here pass cleanly. Whisperstream types the characters into the focused field and uses Right Shift for push-to-talk, so neither axis runs into the screen reader's reserved keys.

None of this means clipboard apps are doomed. Wispr Flow pastes through the clipboard and passed all three screen readers in our testing, and it publishes its own note saying it is tested with NVDA and JAWS. The lesson is narrower: the specific keys an app uses, for both its hotkey and its paste, decide whether a screen reader gets in the way.

Pricing

$29 once (30-day refund, free trial)

Platform

Windows 10/11 x64

Screen readers

Works with NVDA, JAWS, and Narrator (tested)

Trade-off

Windows only; single-user app

• On-device transcription via NVIDIA Parakeet TDT v3 (ONNX, CPU). Your audio never leaves your machine, and there is no cloud transcription path.
• The default push-to-talk key is Right Shift, which NVDA, JAWS, and Narrator do not reserve, so the hotkey fires while a screen reader is on.
• Dictated text is typed into the focused field with a method that does not collide with the screen reader's Insert key, so it lands in Word, your browser, your IDE, and any text box.
• $29 one-time with a 30-day refund and a free trial on first install.
• A searchable, encrypted-at-rest dictation history with audio playback, never synced to a server.

Whisperstream is our pick for a screen-reader user who wants dictation that stays on the PC and costs once. It passed both tests we run, the hotkey fires and the text lands, on NVDA, JAWS, and Narrator. It is not the only app that works, but it is the one that pairs that with fully on-device transcription and a one-time price. For the head-to-head with the best-known cloud option, see Whisperstream vs Wispr Flow.

Pricing

Free, built into Windows

Platform

Windows 10/11

Screen readers

Voice Access co-operates with Narrator (documented)

Trade-off

Win+H is cloud-based; Voice Access covers fewer languages

• Already on every Windows PC, with nothing to install.
• Voice Access (Windows 11) is the newer accessibility feature. It runs on-device after you download a language pack, and Microsoft documents that Narrator users can dictate and edit text with it.
• Voice Typing (Win+H) is the cloud-based one. It uses online speech recognition powered by Azure, so audio leaves your device and an internet connection is required.
• One documented snag: the Alt plus Win plus H auto-punctuation settings dialog does not open under JAWS. That is a settings dialog, not dictation itself.

For a free option that is already on the machine, Voice Access is the honest baseline for screen-reader users: it is built for accessibility, runs on-device, and is designed to work with Narrator. It is worth keeping the two Microsoft features straight. Voice Typing (Win+H) is the cloud-based one. Voice Access is the newer accessibility feature; it can run on-device once you download a language pack. For the side-by-side and the common fixes, see Whisperstream vs Win+H and Win+H not working.

Pricing

$144 / yr (Pro, billed annually); free 2,000 words/week

Platform

Mac, Windows, iOS, Android

Screen readers

Passed all three; Wispr also publishes its own NVDA and JAWS testing

Trade-off

Cloud-only transcription; subscription

• Passed our NVDA, JAWS, and Narrator tests on Windows, on both the hotkey and the text injection.
• Wispr publishes its own accessibility note: it says Flow is tested with VoiceOver, NVDA, JAWS, and TalkBack. Narrator is not named, and we found it works there too.
• Pastes through the clipboard (Ctrl plus V on Windows) and restores your previous clipboard contents afterward.
• Cloud transcription: audio is processed on Wispr's servers, with no on-device mode.
• Pro is $12 per user per month billed annually ($144 a year), or $15 month to month.

Wispr Flow is the strongest cloud pick, and it does not hide from accessibility: it documents its own NVDA and JAWS testing. If you split your day across a Mac, a PC, and a phone, the cross-platform sync is genuinely useful. The trade is the familiar one: a subscription instead of a one-time price, and a cloud pipeline where your audio leaves the device. See Whisperstream vs Wispr Flow for the full comparison.

Pricing

$144 / yr (Pro, billed annually); free 2,000 words/week

Platform

Mac, Windows, iPhone

Screen readers

Passed all three; writes directly at the cursor, no clipboard step

Trade-off

Cloud; subscription; no published accessibility note

• Passed our NVDA, JAWS, and Narrator tests on Windows.
• Inserts text directly at your cursor with no clipboard step, which is part of why it sits cleanly with a screen reader running.
• Added Windows support in early 2026; also on Mac and iPhone.
• Free plan is 2,000 words a week; Pro is $12 a month billed annually ($15 month to month).

Willow Voice passed cleanly on all three screen readers and inserts text right at the cursor, so there is no clipboard chord to collide with. It does not publish an accessibility statement, so our result is first-party. Like Wispr, it is cloud and subscription. For the direct comparison, see Whisperstream vs Willow Voice.

Pricing

Free local tier; Pro from $99.99 / yr

Platform

Mac, iPhone, Windows

Screen readers

Passed all three; free local models, no account

Trade-off

Injection method undocumented; Pro is a subscription

• Passed our NVDA, JAWS, and Narrator tests on Windows.
• Runs local Whisper and Parakeet models on-device for free, with no account, the same engine family Whisperstream uses.
• Genuinely cross-platform: a real Windows 10/11 build, not a Mac-only app.
• The Pro cloud tier is a subscription (about $9.99 a month, or $99.99 a year).

Spokenly is the closest neighbor to Whisperstream in spirit: it runs local models on-device, free, with no account, and it passed all three screen readers in our testing. The paid tier is a cloud subscription, but you do not need it for local dictation.

Pricing

Free and open source (MIT); optional paid Pro

Platform

Windows, macOS, Linux

Screen readers

Passed all three; open source, and clipboard paste still worked

Trade-off

Pastes via the clipboard; more setup than a packaged app

• Passed our NVDA, JAWS, and Narrator tests on Windows.
• MIT-licensed and free. Runs NVIDIA Parakeet on-device, with an optional cloud mode using your own API keys.
• Injects text by automatic pasting through the clipboard, which passed cleanly in our testing.
• Cross-platform across Windows, macOS, and Linux.

OpenWhispr is the free, open-source pick that passed every screen reader we tried. It pastes through the clipboard and still works, which is a good reminder that clipboard apps are not the problem; the specific keys are. See Whisperstream vs OpenWhispr for where a packaged paid app pulls ahead.

Pricing

$96 / yr (Pro, billed annually); free 1,000 words

Platform

Mac, Windows

Screen readers

Text lands, but the default hotkey does not work with Narrator (tested)

Trade-off

Cloud-only; the Narrator hotkey snag

• The typing worked with all three screen readers in our testing: dictated text landed in the focused field.
• But its default activation shortcut did not work with Windows Narrator on, so you may not be able to start dictation with Narrator running until you set a custom hotkey.
• Cloud-only: audio leaves your device, with no offline mode.
• Pro is $8 a month billed annually ($96 a year), or $10 month to month; free tier is 1,000 words.

Aqua Voice is the clearest example on this page that screen-reader support is two questions, not one. Its text injection passed everywhere, but its default hotkey would not fire with Narrator on. If you want to try it with a screen reader, change the activation shortcut first. See Whisperstream vs Aqua Voice for the broader comparison.

Pricing

Free tier; $84.99 / yr or $249.99 once

Platform

Mac, Windows, iOS

Screen readers

Failed all three in our testing

Trade-off

Windows build trails the Mac build

• In our testing it did not work with JAWS, NVDA, or Narrator: neither the hotkey nor the dictated text came through reliably with a screen reader on.
• The Windows build (version 1.x) trails the Mac build (version 2.x), and on Windows that gap shows.
• Runs on-device models; offers a free tier, a subscription ($84.99 a year), and a $249.99 lifetime option.

We are reporting what we observed, not diagnosing Superwhisper's internals: with each of the three screen readers running, we could not get it working. The Windows app already trails the Mac app on version, and for a Windows screen-reader user it is not the pick today. See Superwhisper for Windows for the wider alternatives.

Pricing

Contact sales (Nuance no longer lists a price)

Platform

Windows

Screen readers

Works with screen readers like JAWS (documented)

Trade-off

Expensive and heavyweight; aging desktop product

• The long-standing choice for blind professionals. Nuance says Dragon works with popular screen readers like JAWS.
• A separate paid bridge, J-Say, fuses Dragon and JAWS for full hands-free, non-visual control, and requires an existing JAWS license.
• Windows desktop product. Nuance (now Microsoft) no longer publishes a price; it historically cost several hundred dollars.

If you need deep, professional-grade non-visual control and have the budget, Dragon with JAWS, or the dedicated J-Say bridge, remains the heavyweight standard for blind professionals. For most people the lighter on-device picks above are enough, at a fraction of the cost and setup. It sits last on this list because it fits the fewest people, not because it failed a test. See our Dragon alternatives for the modern local angle.

We tested on Windows 11 with three screen readers, NVDA, JAWS, and Windows Narrator, one at a time. For each app we checked two things: whether the app's activation hotkey fires while the screen reader is running, and whether the dictated text actually lands in the focused field. Both have to work for the app to be usable when you cannot see the screen, so we report them separately.

The seven apps marked Tested in the table we ran ourselves. The three marked Documented (Microsoft Voice Access, Windows Voice Typing, and Dragon Professional) we did not independently test with each screen reader; those rows summarize the vendors' own published support pages, which are linked in Sources. Where a price is billed annually we say so, and where a vendor no longer lists a price (Dragon) we say that too.

Screen-reader behavior can shift between app versions and screen-reader versions. If you depend on a specific app-and-screen-reader combination, try it during a refund window or on a free tier before you commit.

If you want private, on-device dictation that costs once and clears both the hotkey and the injection test, Whisperstream is the direct fit. If you want a free option that is already on your PC, Microsoft Voice Access runs on-device and is built to work with Narrator. If you split your day across a Mac and a phone, Wispr Flow and Willow Voice both passed our tests, at a subscription. If you want a free local tool, Spokenly and OpenWhispr passed too. Avoid Superwhisper for now if you use a screen reader, and if you try Aqua Voice, change its default hotkey before you judge it. For deep, professional non-visual control, Dragon with JAWS, or the J-Say bridge, remains the heavyweight standard.

For the broader picture, see our roundup of the best dictation software for Windows and the head-to-head Whisperstream vs Wispr Flow.