Files
“Clean up my desktop — screenshots into one folder.”
A private voice agent for your Mac
Talk to your Mac. It gets things done.
Ask it to sort your files, draft a reply, or run your tools — it picks the right tool and does it. And when you just want text, dictation lands at your cursor in half a second — with your own Gemini key (BYOK), faster than Wispr Flow.
macOS 14 and later. Apple silicon.
~0.5s latency Unlimited words Voice stays in the EU Native menu-bar app Multilingual Push-to-talk hotkeys
The agent mode
Just say it.
Real commands, spoken the way you’d say them. The agent picks its tools and does the rest — you watch it work, step by step.
Files
“Open the latest invoice PDF and tell me the total.”
Calendar
“What’s on my calendar tomorrow?”
Mail
“Summarize this page and draft an email to the team.”
Mac control
“Close the browser and open Spotify.”
Web
“What does a Mac Studio cost right now?”
Calendar
“Remind me Friday at 9 to call Ronny.”
MCP
“Create a GitHub issue from my clipboard.”
Extend it with MCP servers.
Connect any Model Context Protocol server — GitHub, Notion, databases, browsers, thousands more — and their tools become the agent’s tools. One name, one start command, done.
The cockpit: chat with your Mac.
A floating glass window where you type or speak to the agent. Answers stream in live, work steps stay visible, files attach by drag & drop, and every chat is saved.
Every capability is opt-in and off by default. Anything that changes your Mac asks first — unless you explicitly hand over the keys.
Five things nothing else does.
The reasons people switch and stay.
Flagship
Voice editing
Select any text, hold the agent key, and say “make it more formal” or “fix this”. lisper rewrites it in place.
“make it more formal” replaced in place
Translate while you dictate
Say “translate to English” and keep talking. You get idiomatic translation, not a literal word swap.
Vertical profiles
Medicine, law, developer. Each profile loads its own vocabulary, formatting, and tone.
Spoken snippets
Say “sign-off” and your saved text block drops in, word for word.
Remembers context
Ask a follow-up like “now make it shorter” and the agent remembers the last few minutes of the conversation.
Dictation, done right.
Not a transcript. Clean, formatted text in your voice, with the filler words gone.
um so i basically need to send the the email to the team about the launch being pushed to thursday and tell them to prep the demo
I need to send the email to the team about the launch being pushed to Thursday, and tell them to prep the demo.
Six writing styles
Standard Natural Short Casual lowercase Refined Eloquent Business Bullet points Verbatim
- Spoken formatting like “new paragraph” or “bullet list”
- Vocabulary biasing for names and jargon
- Self-corrections applied as you speak
- Silence gate and hallucination filter: no junk when you pause
Speed comparison
Time from releasing the key to text at your cursor. Lower is better.
lisper Fastest 0.5s
Sprecho 1.3s
Wispr Flow 1.5s
Measured end to end, warm path — lisper with BYOK + Gemini Flash Lite.
Make it yours.
Fifteen voice-reactive HUD styles, your color, your size. The orb glows blue for dictation, purple for prompts, red for the agent.
Speaks your language.
Dictation, the agent, and the whole interface, in many languages, with native recognition.
English Deutsch Français Español 日本語 中文 and more
Why trust lisper
Built in the EU. Your voice never leaves it.
No data broker, no fine print. In GDPR mode — the default — speech recognition and cleanup stay on EU servers.
I wanted Wispr-Flow speed without shipping my voice to a server in the US. So I built lisper: dictation that's GDPR-compliant by default — your audio is processed only on EU servers, in real time, and never stored.
Pierre
Maker of lisper
GDPR by default
In GDPR mode (the default), speech recognition and cleanup run only on EU servers.
Processed, then gone
Your audio is transcribed in real time and discarded. It isn't stored or sold.
24 hours free, cancel anytime
Try everything for a day. Then €9/month, or €6.75 on the yearly plan. Cancel in two clicks.
Simple pricing.
24 hours free, no credit card. Then keep going for less than a coffee a month.
Lifetime
€39 once
one-time · bring your own keys (BYOK, e.g. Gemini) — the 0.5s path
Download for macOS - Unlimited dictation, no word limit
- GDPR by default, voice stays in the EU
- Voice editing, agent mode and prompt mode
- All writing styles and vertical profiles
- 15 voice-reactive HUD styles
- Free auto-updates
The free trial starts when you open the app. Subscribe in-app whenever you are ready.
Questions, answered.
Which platforms are supported?
lisper is a native macOS app today (macOS 14 and later, Apple silicon). More platforms are on the roadmap.
How fast is it really?
About half a second from releasing the key to text at your cursor — with your own Gemini key (BYOK). Clearly faster than Wispr Flow and Sprecho in our tests.
Is there a word limit?
No. Dictate as much as you want, there is no cap and no per-word metering.
What happens to my voice?
Your audio is processed in the EU, GDPR-compliant by default. It is transcribed on the fly and never stored, and your token is bound to your device.
Which languages does it handle?
Dictation and the whole interface work in many languages, with native recognition.
Stop typing.
Download lisper and speak your first sentence in under a minute.
Questions or feedback? Write to hello@lisper.io