Honest ranked list

The best dictation software for Windows in 2026

The Windows dictation market is thinner than the Mac's, and the tool most people reach for — Win+H — stops working without internet. Ten options worth knowing, ranked, with June 2026 prices and honest weaknesses, including our own.

The landscape

Why Windows is the harder platform for dictation

On a Mac you can pick from a crowded field of polished offline dictation apps (we ranked those too: best dictation software for Mac). On Windows, the built-in tool everyone knows, Voice Typing (Win+H), sends your audio to Microsoft's cloud and refuses to run offline; the built-in that is offline, Voice Access, is an accessibility command tool that never shipped for Windows 10.

Beyond the built-ins: a $699 legacy heavyweight (Dragon), cloud subscriptions ($96–180 per year, forever), and free open-source tools that expect you to pick engines and download models yourself. That gap — offline, polished, paid once — is the one we built SuperDictate to fill.

1 — Best for private, everyday dictation you pay for once

SuperDictate

Hold a hotkey, talk, and SuperDictate types where your cursor is — any app, Windows 10 or 11. Audio is processed 100% on-device and never uploaded; it works in airplane mode. The speech model runs on your device. Optional AI polish is text-only, off by default, via local Ollama or your own API key. Pay $79 once, own it forever — one license covers your Mac and your Windows PC. Free to try — 15 minutes every month, every feature, no credit card — with a 30-day money-back guarantee.

Honest weaknesses: we're newer, with a smaller community — Dragon has a 25-year ecosystem of forums, consultants, and specialized vocabularies we can't match. SuperDictate dictates; it doesn't drive your PC by voice — for that, look at Voice Access or Talon. And the free tier is 15 minutes a month: enough to evaluate, not enough to live on.

2 — Best free option, if you're always online

Windows Voice Typing (Win+H)

Free, zero install, one keystroke away in any text box on Windows 10 and 11, with 40+ languages and spoken punctuation. Recognition runs on Azure Speech, which is also the catch: Microsoft's own support page is blunt that voice typing needs an internet connection. Your audio is processed on Microsoft's servers, and users on Microsoft's own Q&A forums report "no internet" errors even while connected. Fluid dictation (Copilot+ PCs only) fixes grammar and filler words on-device, but it's a correction layer — recognition itself still goes through the cloud.

3 — Best built-in offline option (Windows 11 only)

Windows Voice Access

The opposite of Win+H: Microsoft states Voice Access works without an internet connection, with on-device recognition after a one-time language-file download. Free, genuinely private, and it controls the whole PC — open apps, click buttons, dictate with corrections. The trade-offs: Windows 11 22H2 or later only (Windows 10 users get nothing), roughly 15 languages versus 40+ for voice typing, and it's a command-and-control accessibility tool first, so prose dictation feels clunkier than a dedicated app.

4 — Best for correction-by-voice and specialized vocabularies

Dragon Professional v16

Still the benchmark for deep voice workflows: correct-by-voice editing, custom vocabularies and commands, transcription of recordings, and a 25-year ecosystem strong in legal and professional niches — and it's a perpetual license, which we respect. But it costs $699 for one user (Staples retail, verified June 2026; Nuance lists no public price), it's Windows-only since the Mac version was discontinued in 2018, and no new desktop version has shipped since v16 in 2023. If you don't need the command depth, you're paying nine times our price — full breakdown in our Dragon alternative guide.

5 — Best subscription app, if cloud processing is acceptable

Wispr Flow

The most polished of the modern cloud tools: fast, auto-edits filler and formatting, 100+ languages, SOC 2 Type II, on Windows, Mac, iPhone, and Android. As of June 2026: $15/month, or $12/month billed annually (about $144/year), with a 2,000-words-per-week free tier and a 14-day Pro trial. Two structural problems: the subscription never ends, and there's no offline mode — Wispr's own privacy page says transcription always happens in the cloud. Privacy Mode controls retention, not where your audio is processed. Full comparison: SuperDictate vs Wispr Flow.

6 — Best free open-source offline tool

Handy

Free, MIT-licensed, and completely offline on Windows, macOS, and Linux — its README promises your voice stays on your computer, and it delivers. Push-to-talk hotkey, winget install, and a choice of open-source models: Whisper on GPU or Parakeet V3 on CPU. As of spring 2026 it had roughly 20,000 GitHub stars and active releases — credible, maintained software. It's also deliberately minimal: no AI cleanup, no formatting profiles, no commercial support, and you download models yourself. If that sounds fine, Handy is the best free thing on this list.

7 — Best for tinkerers who want to choose everything

Whispering (Epicenter)

A free, open-source, ~22MB desktop app for Windows, Mac, and Linux where every layer is your choice: run a fully local engine (no internet, no API keys) or bring your own cloud key from Groq, OpenAI, or ElevenLabs and pay the provider directly. Data stays local either way, and the code is auditable. The cost is configuration: engines, models, and API keys are decisions mainstream users won't make, and there's no support contract when something breaks.

8 — Best for voice coding, not prose

Talon

Free for Windows, Mac, and Linux, with a local Conformer speech engine, Python scripting, custom command grammars, noise input, and eye tracking. For hands-free computing — especially voice coding — nothing else here comes close. But Talon is built for commands and code: plain dictation is a secondary mode, the learning curve is steep, and there's no commercial support. A Patreon tier offers beta builds — historically around $15–25/month, though we couldn't re-verify the current price as of June 2026.

Also worth knowing

Voicy and Aqua Voice

Voicy is the budget commercial option: $8.49/month billed annually or $260 lifetime as of June 2026, on Windows, Mac, Linux, and Chrome. Two cautions: the free trial is 30 minutes total, ever — not recurring — and while Voicy says transcripts are stored only on your device, its site doesn't document where recognition actually runs, so we won't call it offline.

Aqua Voice is fast, low-latency cloud dictation on its hosted Avalon model: Pro is $8/month billed annually, with a free tier of 1,000 words total — not recurring. Windows is supported, but as of June 2026 the site leads with Mac and iOS downloads. Both are subscription-first; neither is verifiably on-device.

Questions

Windows dictation, answered