DictationRank
Which language do you dictate in?
Dictation accuracy varies enormously by language. Pick yours and we'll re-rank every app for it.
The price landscape
Every ranked product, plotted by headline price. Switch between recurring and pay-once pricing — the two halves of this market barely overlap.
Pricing clusters into four bands: free and open source, consumer subscriptions at roughly $5–15 a month, one-time licences between about $49 and $99, and the professional tier where Dragon still asks around $630 for a perpetual licence. Several products appear in both views — Superwhisper and Voicetypr sell a subscription and a pay-once option.
What do you dictate all day?
Tap anything that applies. Rankings re-weight toward apps that handle that kind of text well.
Where should your audio go?
The main fault line in this market. Apps built for a different stance drop down the ranking.
All dictation apps, scored
15 apps ranked for Español.
Wispr Flow
The product that turned voice typing into a category. Wispr Flow listens on a held key, cleans up filler and false starts with its own model, and pastes finished text into whatever app has focus. It is also cloud-only, which is exactly the opening every challenger on this page aims at.
Superwhisper
Runs Whisper models entirely on your machine, then optionally passes the transcript through a frontier LLM you supply keys for. Modes let you define per-app behaviour: raw transcript in the terminal, polished prose in your editor, a bulleted summary in notes.
MacWhisper
Jordi Bruin's Whisper front-end grew from a weekend utility into one of the best-selling indie Mac apps ever, with more than half a million sales on Gumroad. Pay once, transcribe locally forever — dictation, meeting recordings and dropped media files in the same window.
Aqua Voice
A Y Combinator W24 company that abandoned Whisper for its own Avalon model, chasing sub-second turnaround. The hold-to-talk interaction is the tightest here: text appears about as fast as you finish the sentence.
Handy
MIT-licensed, cross-platform, fully local: Whisper and NVIDIA Parakeet V3 both run on-device, with a push-to-talk shortcut that types into any application. The closest an open project comes to matching the commercial leader's daily feel.
Willow Voice
The closest functional match to Wispr Flow, differentiated by compliance: a business tier with an enforced HIPAA posture and signed agreements, which matters if you dictate anything clinical.
VoiceInk
GPL-licensed with $29–69 prebuilt binaries, and unusually polished for an open project: on-device transcription, per-app profiles and a dictionary, with none of it phoning home.
OpenWhispr
An MIT project that says the quiet part out loud: it exists to be the free open-source WisprFlow alternative. Local Whisper by default, optional cloud API keys if you want a stronger model.
Voicetypr
AGPL source you can build yourself, or a $69 one-time binary if you would rather not. The pitch is explicitly political: dictation should not be a subscription, and your audio should not leave the machine.
WhisperTyping
Windows-only by choice, with tiers from $5 to $30 a month including a medical plan. It gets into the awkward corners of Windows — old line-of-business apps, remote desktop sessions — where Mac-first tools never go.
SpeechPulse
A pay-once offline dictation tool that leans on your GPU to make large local Whisper models fast. $99 per operating system, no subscription, no telemetry, no login.
VoiceOS
Goes past transcription into agentic territory: an Agent Mode that takes spoken instructions and acts on the app in front of you — rewriting a selection, filling a form, reorganising a document — rather than just typing what you said.
Monologue
Bundled with Every and positioned bluntly as a Wispr Flow replacement. Its bias is long-form: it holds context across paragraphs, keeps your voice rather than flattening it, and behaves well when you ramble and then correct yourself.
FreeFlow
A tiny MIT Mac app billed as a free and fast alternative to Wispr Flow. It skips local models entirely and sends audio to Groq with your own key, which makes it startlingly quick for roughly cents a day.
Dragon Professional v16
The product everyone else is reacting to. A perpetual licence around £499 through resellers, deep custom-command scripting, and an offline engine — while the healthcare line has been folded into Microsoft's Dragon Copilot.
Quick comparison
| App | Price | Languages | Processing | Offline capable | Speech engine | Best for |
|---|---|---|---|---|---|---|
| Wispr Flow | Free 2,000 words/week · Pro $15/mo | 100+ | Cloud only — no offline mode | — | Proprietary (Canto) speech + cleanup model | The category leader |
| Superwhisper | Free tier · Pro $8.49/mo · $249 lifetime | 100+ (local models) | Fully on-device by default | — | Local Whisper / Parakeet + optional BYO LLM | Best privacy-first pick |
| MacWhisper | Free tier · Pro €64 one-time (~$69) | 100+ (local models) | Fully on-device | — | whisper.cpp with large-v3 and Parakeet models | Best one-time purchase |
| Aqua Voice | Free 1,000 words · Pro $10/mo | 40+ | Cloud only | — | Proprietary Avalon model | Fastest cloud dictation |
| Willow Voice | Free tier · Pro $15/mo · HIPAA business tier | 50+ | Cloud, with compliance controls | — | Proprietary cloud model | Best for regulated work |
| VoiceOS | Free tier · Pro $11.99/mo | 50+ | Cloud only | — | Cloud speech model plus LLM agent layer | Most ambitious |
| Monologue | Pro $15/mo (bundled with Every) | 40+ | Cloud only | — | Cloud speech model plus editorial rewrite pass | Best for long-form |
| WhisperTyping | Free tier · $5 / $15 / $30 per month | 60+ | Cloud, medical tier with stricter handling | — | Whisper-family cloud models | Cheapest subscription |
| SpeechPulse | $99 one-time per OS | 90+ (local models) | Fully on-device | — | GPU-accelerated Whisper / Parakeet | Offline, pay once |
| Dragon Professional v16 | ~$630 perpetual licence | 6 core languages | On-device engine | — | Nuance legacy engine with speaker profiles | The legacy incumbent |
| Handy | Free (MIT) | 100+ (Whisper) / 25 (Parakeet) | Fully on-device | — | whisper.cpp + Parakeet V3 | Best free alternative |
| OpenWhispr | Free (MIT) | 100+ (local models) | Fully on-device, optional cloud | — | Local Whisper with optional API backends | Free clone of the leader |
| FreeFlow | Free (MIT) · Groq usage at cost | 90+ (Whisper on Groq) | Cloud via your own Groq key | — | Whisper large on Groq inference | Fastest free option |
| Voicetypr | Free from source · $69 one-time binary | 100+ (local models) | Fully on-device | — | Local Whisper / Parakeet | Anti-subscription |
| VoiceInk | Free from source · $29–69 one-time | 100+ (local models) | 100% on-device | — | whisper.cpp with Parakeet option | Best hybrid model |
Best desktop dictation apps in 2026: the market, mapped
Twenty-seven products were verified for this report and fifteen are ranked here: ten commercial and five open source. The market split cleanly along two lines — where your audio is processed, and whether you pay monthly or once.
Pricing clusters into four bands: free and open source, consumer subscriptions at $5–15 a month, one-time licences at roughly $49–99, and a professional tier where Dragon still asks around $630. Two technical shifts matter: NVIDIA's Parakeet V3 made fast on-device transcription practical on ordinary hardware, and LLM cleanup turned raw transcripts into finished text. The main attack axis on the leader is privacy — almost every challenger sells on-device processing that Wispr Flow structurally cannot offer.
Wispr Flow still produces the cleanest finished text of anything tested — at $15 a month, with every word processed in the cloud.
Superwhisper runs Whisper locally and only calls out to an LLM when you ask it to, for about half the leader's price.
Handy is MIT-licensed, cross-platform including Linux, and fast enough on Parakeet V3 to replace a subscription for most people.
MacWhisper and VoiceInk both trade the monthly bill for a one-time licence and keep transcription entirely on the machine.
Dragon Professional survives on voice control and macros rather than transcript quality — open models overtook its engine years ago, but nothing matches its scripting for hands-free computing.
How dictation apps work
Plain-language explainers for the technology behind the rankings.
How language support really works in dictation apps: why so many apps share the same 99-language list, what Parakeet V3 changed for European languages, and how to tell real support from a marketing count.
A transparent look at DictationRank's ranking model: the hard gates (platform and privacy stance), the weighted scores across five use cases and per-language accuracy, and a worked example.
The popularity league table for desktop dictation, built from the only public signals that exist: App Store rating counts, GitHub stars, and vendor-published user numbers.
Which dictation apps handle Spanish properly: inverted punctuation, regional accents, tuteo vs usted register, and Spanish-English code-switching, tested across the full ranking.