Skip to content
Dictaur
100% on-device · free

Dictate anywhere. Nothing leaves your machine.

On-device AI dictation for Linux, macOS & Windows. Press a hotkey, speak, and your words appear in any app. Transcription is free and 100% local — add hosted AI only if you want it.

  • No account to dictate
  • Linux · macOS · Windows
  • 99+ languages
dictaur — on-device

$ press ⌘⇧Space · speak…

🎙 whisper running locally (no upload)

↳ "ship the build and write the release notes"

✓ text inserted at cursor — 0 bytes of audio sent

How it works

Hotkey. Speak. Done.

No window to open, no upload, no waiting on a server.

  1. Step 1

    Press your hotkey

    A global shortcut starts listening — from any app, anywhere on your desktop.

  2. Step 2

    Speak naturally

    Whisper transcribes on your device, in real time, in 99+ languages.

  3. Step 3

    Text appears at your cursor

    Your words land in the editor, browser, chat box, or terminal you're already in.

100% on-device, free. No account, no upload, no subscription to dictate.

Your voice never touches a server.

Cloud dictation makes you subscribe just to turn speech into text — and uploads your voice to do it. Dictaur runs the model on your machine: your audio and transcripts never leave your device, and there's no account to transcribe. Dictation is free; you only pay if you want the hosted-AI convenience.

Why Dictaur

Local-first, cross-platform, honest pricing

Local & private

Whisper runs on your device. Your audio and text never leave your machine. No account needed to transcribe.

Linux, macOS & Windows

The only polished dictation app that treats Linux first-class — .deb + .AppImage, X11 recommended (Wayland supported, with some compositor limits) — alongside macOS and Windows.

Free, local core

Dictation is free and runs entirely on your machine. No subscription just to put speech into text.

Optional hosted AI

Want polish, summaries, translation & chat with zero setup? Dictaur Pro, $10/mo — no API key. Or bring your own key for free.

Works in any app

A global hotkey inserts text at your cursor — editors, browsers, chat, ChatGPT, Cursor, your terminal, anything.

99+ languages

Whisper auto-detects and transcribes 99+ languages. Pick a tiny model for speed or large-v3 for accuracy.

Transcription is free & local. Add hosted AI only if you want it.

Want polish, summaries, translation, and chat with zero setup — no API key, no Ollama to run? Dictaur Pro is $10/mo, optional, and only runs when you ask. Prefer to stay 100% free? Bring your own API key.

How it compares

The incumbents make you subscribe just to dictate

Dictaur lets you dictate free and locally — and only charges for the optional AI convenience.

CapabilityDictaurWispr FlowSuperwhisper
LinuxYesNoNo
macOS & WindowsYesYesYes
On-device transcriptionYesNoPartial
Free local coreYesNoPartial
Pay only for cloud AIYesNoNo
Hosted AI, no API keyOptionalForcedYes

Verifiable privacy

Trust the architecture, not a promise

Your audio and text never leave your device. The only things that touch the network are updates, the one-time model download, your account, and hosted AI — only if you turn it on.

Audio & text stay local

Transcription is processed on-device. Your raw audio and transcripts never leave your machine — survivable under a packet-capture challenge.

No content telemetry

The app ships zero analytics. It never phones home with what you said or typed. (This marketing site measures its own funnel — the app does not.)

Signed & notarized builds

Installers are code-signed, and the macOS build is notarized. You run what we shipped.

Cancel anytime

Dictaur Pro is month-to-month with a 7-day money-back guarantee. The free local app keeps working regardless.

Download

Get Dictaur — free

Install the free local app. No account required to dictate.

Pricing

Free to dictate. Pay only for hosted AI.

Wispr Flow is $15/mo and uploads your voice. With Dictaur, transcription is free and 100% local — hosted AI is optional.

Free

$0

Local transcription, on your device.

  • Unlimited offline transcription
  • All exports, 99+ languages
  • Bring your own API key / local Ollama for AI
  • No account required
Download — free
Most popular

Pro

$10/mo

Hosted AI, zero setup.

  • Everything in Free
  • Hosted AI: polish, summarize, translate, chat
  • No API key — fair-use quota
  • $96/yr billed annually
Get Dictaur Pro

Power

$18/mo

For heavy users.

  • Everything in Pro
  • Higher limits, live polish on streaming
  • Voice-call minutes, premium TTS
  • $168/yr billed annually
Get Power

Team

$15/seat/mo

3-seat minimum.

  • Pro for every seat
  • Shared vocabulary
  • Admin & central billing
Start a team

Cancel anytime; 7-day money-back. Dictaur Pro is $10/mo. The free local app is free forever.

FAQ

Questions, answered

Is it really private?

Yes. Transcription runs on-device via Whisper. Your audio and text never leave your machine, and you don't need an account to dictate.

Is it free?

The app and local transcription are free. Dictaur Pro ($10/mo) is optional and only adds the hosted-AI features.

What's the difference between Free and Pro?

Free = local dictation plus bring-your-own AI API key. Pro = hosted AI (polish, summaries, translation, chat) with no key and no setup.

Does it work offline?

Dictation works completely offline after the one-time model download. The hosted-AI features need the internet, and they're opt-in.

Does it really run on Linux?

Yes — .deb and .AppImage, Linux is first-class alongside macOS and Windows. X11 is the recommended session (global hotkeys and text insertion work best there); Wayland is supported but limited by compositor restrictions.

Is my voice sent anywhere?

No. Audio and transcripts are processed on-device and never leave your machine. The network is only touched for updates, the model download, your account, and hosted AI if you turn it on — never your raw audio.

Can I cancel?

Anytime, with a 7-day money-back guarantee. The free local app keeps working regardless.

Which languages are supported?

99+, auto-detected by Whisper. Pick a tiny model for speed or large-v3 for accuracy.