Press ⌥ Space
In any app, mid-sentence. A discreet capsule appears to confirm listening has started. You can also hold the key to talk for as long as it is pressed.
Voice dictation for macOS, 100% on device
Dictate anywhere. Everything stays on your Mac.
One shortcut, you speak, and the text is written at your cursor in any application. Recognition runs on the Apple chip: not a sound, not a word leaves your computer.
Full trial, no cardLifetime license €79or €6.99/monthmacOS 26 · Apple silicon
To:Claire Dupont
Subject:Notes from Monday’s meeting
Hi Claire, here are the three decisions from Monday: the launch is confirmed for the 12th, the mock-up goes to review on Thursday, and I’ll send you the revised budget
How it works
Three moves, and it’s written
Speak naturally
Provisional text appears in the capsule as you talk: confirmed in white, in progress in grey. Say “new line” or “new paragraph” to structure your text.
Press again
The final text, punctuated and optionally cleaned of filler words, is inserted in one go where your cursor was. Esc cancels at any time.
Features
A complete dictation tool, built with care
Designed for French first, and for a couple of dozen other languages. Everything works offline once the models are downloaded.
Text lands where your cursor is
Mail, Notes, Pages, Slack, a browser, a terminal: one shortcut, you speak, the text is inserted when you stop. Nothing to copy and paste.
Fast, even on long dictations
Whisper runs on the Apple chip; provisional text shows while you speak. Parakeet, optional, is faster still for 25 European languages.
A profile for every app
Language, AI mode, translation, filler-word cleanup and insertion method, set per application. Ready-made profiles cover the common cases.
Careful punctuation, French included
Proper French spacing and quotation marks, times and addresses left intact. Voice commands such as “new line”, “new paragraph” or “undo that” are understood.
AI modes, on device
Fix, Clear and concise, Professional tone, Email, Bullet list, Summary, Prompt for an AI, Code prompt: formatting runs on the macOS language model, on your Mac.
Dictate and translate
Speak French, get English, Spanish, German… Translation uses the macOS system framework and stays local.
Subtitles from files
Drop an audio or video file: DictateLocal writes an SRT or VTT file, with speaker identification if you want it.
Live captions
A call in a foreign language, a podcast, a video: the Mac’s audio is captioned at the bottom of the screen, translated on request.
Meeting mode with a summary
Your microphone and the Mac’s audio are transcribed side by side, speakers are identified, and a summary is written, all without leaving the machine.
Interface in ten languages
French, English, Spanish, German, Italian, Portuguese, Dutch, Arabic, Japanese and Simplified Chinese.
Shortcuts your way
Short press to start and stop, hold to talk, or a single key (right ⌘, fn…). Pause, cancel and switch modes from the keyboard.
One-click updates
New versions are offered inside the app, signed and verified, through Sparkle. Updates are included with the license.
Privacy
Nothing leaves the Mac. Verifiably.
DictateLocal has no transcription server, no user account, no usage analytics. The Whisper and Parakeet models run on your machine; your dictation history is encrypted and stored locally, and you can turn it off.
- No audio and no dictated text is ever transmitted.
- Only three connections, all declared: licensing, updates, model downloads.
- Cloud engines (Groq, OpenAI) exist as an option, with your own key and explicit consent; they are never used as an automatic fallback.
- The app’s network log lists every connection it makes, so you can check for yourself.
| Destination | Why | Audio |
|---|---|---|
api.lemonsqueezy.com | License check | never |
dictatelocal.com | Update check | never |
huggingface.co | Model download, once | never |
YouDo we keep the 12th for the launch?
Speaker 2Yes, as long as the mock-up is reviewed on Thursday.
Speaker 3I’ll send the revised budget by tomorrow.
Summary (generated on this Mac)
- Launch confirmed for the 12th.
- Mock-up review on Thursday.
- Revised budget due tomorrow.
Meetings and captions
Beyond dictation
Meeting mode transcribes your microphone (You) and the Mac’s audio (everyone else) side by side, identifies the speakers at the end and writes a local summary. Live captions show what is being said in a call or a video at the bottom of the screen, translated if you like.
On livre la mise à jour jeudi, après la revue.
We’ll ship the update on Thursday, after the review.
Both features need the macOS “Screen & System Audio Recording” permission, used only to capture the Mac’s sound. Nothing is saved to disk unless you ask.
Pricing
Lifetime license or subscription
Same features, same 3 Macs, same updates. You only choose how to pay.
Subscription
€6.99/month
about $7.50 · incl. VAT, no commitment
or €49/year (about $53), two months free
- Every feature, 3 Macs
- Updates included while subscribed
- Cancel any time
Choose monthly or yearly at checkout.
Most popular
Lifetime license
Launch price
€99€79
about $85 · incl. VAT, one-time payment
- Every feature, 3 Macs
- Updates included, no time limit
- 14-day refund
Limited time; regular price €99.
3-day free trial before you buy. Secure checkout by Lemon Squeezy (merchant of record), card or PayPal, VAT and invoice handled automatically. Prices in EUR; your local currency is shown at checkout, and dollar amounts are approximate. Comparison and pricing questions.
FAQ
The short answers
Are my recordings sent anywhere?
No. Audio is captured in memory, transcribed by a model running on your Mac, then released. The app never sends audio or dictated text. The network log, in Settings, About, lists every connection it makes.
Is the free trial limited?
No: three days with every feature, no account and no credit card. After three days the app asks for a license key to keep dictating.
How many Macs can I use my license on?
A license activates on up to three Macs. You can deactivate a Mac from the app (Settings, About, “Deactivate on this Mac”) to free a seat.
What are the requirements?
macOS 26 (Tahoe) or later. A Mac with Apple silicon (M1 or later) is recommended: recognition models run much faster there. Models are downloaded once (from a few hundred MB to a few GB depending on the variant) and then work offline.
Is DictateLocal on the Mac App Store?
No, DictateLocal is sold directly, through Lemon Squeezy. The app is signed and notarized by Apple; updates happen inside the app.
Try DictateLocal for three days
Every feature, no account, no credit card. Download, open the disk image, drag the app to Applications: your first dictation is one shortcut away.
Requirements: macOS 26 (Tahoe) or later · Mac with Apple silicon recommended · Signed and notarized disk image