Voice to text,
anywhere.
WisprType is a native macOS dictation app powered by on-device AI. Hold a key, speak, and your words appear wherever you type — private by default, blazingly fast on Apple Silicon.
- Apple Silicon
- 100% on-device
- Free forever
- No account required
Hey team, just wanted to share a quick update on the progress so far
Hold the hotkey · speak · your words appear
Three steps. No setup required.
The fastest path from thought to text on your Mac.
Hold
Press and hold your hotkey (Right ⌘ by default).
Speak
Talk naturally while the aurora pill glows.
Release
Your words are transcribed and pasted instantly.
Write everywhere you type.
WisprType pastes into any focused text field. No integrations. No plugins.
Email & long replies
Dictate full emails in seconds. Sound natural, not robotic — even on mobile-length thoughts.
Code comments & PRs
Capture intent for reviewers without breaking flow. Works in VS Code, Xcode, and your terminal.
Slack & iMessage
Reply to messages 3× faster than typing. Dictation pastes straight into the focused chat.
Docs & notes
Notion, Apple Notes, Bear, Obsidian — anywhere a cursor blinks, your voice can write.
Forms & search bars
Bug reports, support tickets, address fields — fill in any text input by holding a key.
Terminal commands
Speak the long flag-heavy commands you forget. Edit and re-run without leaving the shell.
Speak how you actually speak.
An optional AI pass cleans your dictation as it lands — fillers, stutters and self-corrections vanish, punctuation appears, and what gets typed reads like you wrote it.
Speak.
We polish.
An optional AI pass scrubs filler words, repairs self-corrections, and lands clean punctuation — so what gets typed reads like writing.
um so I was thinking maybe, uh, we should send the email today, no actually first thing tomorrow morning
I was thinking we should send the email first thing tomorrow morning.
Goodbye, fillers.
The little words you say but never want to write.
- um
- uh
- so
- like
- basically
- you know
- yeah
- kinda
Perfect commas, periods, dashes.
what time is the meeting tomorrow
What time is the meeting tomorrow?
Changed your mind? We caught it.
Smart Typing keeps the version you actually meant.
Send the deck today, no actually, tomorrow morning before the standup.
Send the deck tomorrow morning.
Stays private.
Telemetry never includes the words you said or the words we cleaned.
Toggle it any time.
Polish on for high-stakes writing. Off when you want the raw transcript.
Your words, your way.
Teach WisprType the names, jargon, and acronyms you use every day — organized by category so it knows the right meaning, every time.
Names
6 entries- Aarav Patel
- Wispr
- Anthropic
- Substack
- Linear
- Figma
Technical
6 entries- TypeScript
- Kubernetes
- GraphQL
- Postgres
- RAG
- OAuth
Medical
5 entries- myocardial
- ibuprofen
- hypertension
- electrolyte
- ECG
Legal
4 entries- voir dire
- amicus curiae
- force majeure
- prima facie
Get domain-specific words right the first time. No more renaming “Aarav” to “Arav” ten times a day.
Everything you need.
Powerful capabilities packaged in a tiny, focused app.
Runs entirely on your Mac
Audio is transcribed on-device. No internet required, no servers in the loop, no data leaves your machine.
Smart Typing cleanup
Optional AI pass removes filler words, fixes self-corrections, and adds punctuation — your transcript reads like writing.
Custom dictionary
Add names, jargon, and acronyms organized by category so the words that matter to you come out right every time.
Push-to-talk, your key
Pick Right ⌘, Right ⌥, Right ⌃, or the Globe key. Hold to speak, release to insert. Never blocks the key for normal use.
Floating aurora pill
A glassy capsule appears while you dictate, glowing through recording, transcribing, and inserting — then quietly disappears.
Bring your own key
Plug in OpenAI, Groq, or Deepgram for cloud-grade accuracy when you want it. Your API keys stay on your machine.
Always ready
Optionally launches at login and stays out of the way until you press your hotkey. Wakes instantly, every time.
Pastes anywhere
Drops your transcript into the focused app — emails, code editors, chat, browsers, terminals, forms. No integrations needed.
Choose your engine
Five Whisper sizes from Tiny to Large v3 — pick fast or pick accurate, swap any time. Models cache locally on first use.
A glance is all it takes.
A glassy capsule sits above your work and shifts color to show what's happening — recording, transcribing, polishing, inserting. No popups, no focus stealing.
Recording
Capturing what you say in real time.
Auto-cycling · click any phase to focus
Built for the way you actually work.
A clear, honest comparison vs. the alternatives — so you can decide for yourself.
| Capability | WisprType | macOS Dictation | Cloud apps |
|---|---|---|---|
| 100% on-device processing | Partial | ||
| Sub-second latency on Apple Silicon | Partial | ||
| Choose your own model | Partial | ||
| Custom hotkey, push-to-talk | Partial | Partial | |
| No subscription, no account | |||
| Pastes into any focused app | Partial |
100% on-device processing
- WisprType
- macOS
- Partial
- Cloud apps
Sub-second latency on Apple Silicon
- WisprType
- macOS
- Partial
- Cloud apps
Choose your own model
- WisprType
- macOS
- Cloud apps
- Partial
Custom hotkey, push-to-talk
- WisprType
- macOS
- Partial
- Cloud apps
- Partial
No subscription, no account
- WisprType
- macOS
- Cloud apps
Pastes into any focused app
- WisprType
- macOS
- Cloud apps
- Partial
Privacy-first, always.
WisprType runs locally by default. Your audio is processed on-device — no data leaves your machine. Optional anonymous telemetry never includes transcripts, audio, or personal content.
Choose your engine.
Local Whisper models or cloud providers — whichever fits your work.
Local models
DefaultRun entirely on your Mac
- Whisper Tiny~75 MBFastest
- Whisper Base~150 MBRecommended
- Whisper Small~500 MBBetter accuracy
- Whisper Medium~1.5 GBHigh accuracy
- Whisper Large v3~3 GBBest accuracy
Cloud providers
OptionalBring your own API key
- OpenAIgpt-4o-transcribe
- Groqwhisper-large-v3-turbo
- Deepgramnova-3
Your API keys are stored locally and never transmitted to WisprType servers.
Got the download? Here's how to run it.
Drag WisprType into Applications, grant a couple of permissions, and you're dictating in seconds.
Install in 30 seconds
- 01
Open the DMG
Double-click the downloaded file and drag WisprType into your Applications folder.
- 02
Launch WisprType
Open it from Applications, grant Microphone and Accessibility permissions, then hold Right ⌘ to start dictating anywhere.
Common questions.
If you don't see your question, get in touch.
Is WisprType really free?
Yes. WisprType is free to download and free to use. The local models run entirely on your Mac at no cost. If you choose to use a cloud provider (OpenAI, Groq, Deepgram), you bring your own API key and pay them directly — WisprType doesn't take a cut.
What is Smart Typing?
Smart Typing is an optional AI cleanup pass that runs after your transcript is generated. It removes filler words like 'um' and 'uh', fixes self-corrections, and adds punctuation — so what gets typed reads like writing instead of a raw stream of speech. Toggle it on for emails and documents; turn it off when you want the verbatim transcript.
Why does it require Apple Silicon?
WisprType is heavily optimized for the Neural Engine on M-series chips. This is what enables sub-second, fully on-device transcription. Intel Macs aren't supported in v0.1.
Does WisprType send my voice or transcripts to the cloud?
No, not by default. All speech-to-text happens on your device. The only exception is if you explicitly opt into a cloud provider by entering an API key, in which case audio is sent to that provider only — never to WisprType.
Can I teach it the names and jargon I use?
Yes — the Custom Dictionary lets you add words organized by category (Names, Technical, Medical, Legal, Other). Once added, those words come out right every time you dictate, so you stop fixing the same misheard term over and over.
Which model should I use?
Start with Whisper Base (~150 MB). It's the recommended default — fast and accurate enough for most dictation. Step up to Small or Medium for noisier environments or technical vocabulary. Large v3 is the most accurate but uses ~3 GB and is slower on lower-tier chips.
Can I change the hotkey? What if Right ⌘ conflicts with something?
Yes — pick from Right ⌘, Right ⌥, Right ⌃, or the Globe (Fn) key. WisprType uses push-to-talk, so it only listens while the key is held down — it never blocks the key for normal use.
Still wondering? Send us a question.
Start speaking, stop typing.
Download WisprType for free and experience voice-to-text that actually works — fast, private, and always ready.
Requires macOS on Apple Silicon · Free