Voice dictation for Mac & Windows

Talk. Release. Typed.

Hold a hotkey, speak, and your words land in any app. Everything runs on your Mac or PC. No cloud. No subscription.

Runs 100% locally  ·  No account needed  ·  Free to try

Notes
Start typing or hold your hotkey to dictate…
Hold ⌥Space to record
Idle
Recording
Processing
Pasted

How it works

Four steps. That’s it.

01

Hold

Press and hold your global hotkey, WhisperVoca activates instantly, even if it's not in focus.

02

Talk

Speak naturally. The Dictation Overlay shows live audio waves so you know you're being heard.

03

Release

Let go of the key. WhisperVoca transcribes your speech right on your Mac or PC. Nothing gets sent anywhere.

04

Pasted

Your words land in whichever field you were in, Notes, Slack, VS Code, email, anything.

Works everywhere

Dictate into any app you can type in.

Slack, Cursor, Notion, Gmail, VS Code, Teams… WhisperVoca works at the OS level, if it has a text field, you can dictate into it.

Cursor
Claude Code
VS Code
Codex
Amp
GitHub
Copilot
Windsurf
Zed
JetBrains
Replit
Warp
iTerm
Neovim
Cursor
Claude Code
VS Code
Codex
Amp
GitHub
Copilot
Windsurf
Zed
JetBrains
Replit
Warp
iTerm
Neovim
Slack
Teams
Discord
Zoom
Gmail
Outlook
Linear
Notion
Asana
Jira
Trello
Monday
Airtable
Superhuman
Telegram
Slack
Teams
Discord
Zoom
Gmail
Outlook
Linear
Notion
Asana
Jira
Trello
Monday
Airtable
Superhuman
Telegram
Word
Bear
Obsidian
Figma
ChatGPT
Gemini
Google Docs
Apple Notes
Craft
Logseq
Canva
Framer
Ulysses
Roam
Substack
Word
Bear
Obsidian
Figma
ChatGPT
Gemini
Google Docs
Apple Notes
Craft
Logseq
Canva
Framer
Ulysses
Roam
Substack

8 dictation modes

Tuned for how you actually write.

Pick the mode that matches what you're writing. Medical notes come out clinical, Slack messages stay casual, emails read professionally.

💬

General

Natural, neutral English

Default, everyday dictation, quick notes, anything

You say

hey just wanted to check in on the project timeline uh do you have an update

WhisperVoca pastes

Hey, just wanted to check in on the project timeline, do you have an update?

Agentic workflows

Faster agentic coding. Without the keyboard.

Works where you code

Cursor, Claude Code, Codex, Amp. Hold, speak, let go. Your prompt lands straight in the editor.

Less typing, more building

Talking through a change is faster than typing it. The agent gets more context and the result is usually better.

One voice, any window

Whichever window is in focus gets your words. No copy-paste, no switching back and forth.

Cursor

Refactor the auth middleware to support JWT refresh tokens. Add error handling for expired tokens and log failed attempts.

C

I'll refactor the auth middleware. Let me start by updating the token validation logic…

 

```typescript

const verifyToken = async (token: string) => {

try {

return jwt.verify(token, process.env.JWT_SECRET);

} catch (e) {

if (e instanceof TokenExpiredError) {

return refreshToken(token);

}

logger.warn('Auth failure', { token });

throw new AuthError('Invalid token');

}

};

```

Privacy & independence

Your data is 100% yours. Always.

Your voice never leaves your computer. No cloud processing, no account, no server. Just Whisper running locally on your machine.

Voice never leaves your device

All transcription happens locally using Whisper AI. Nothing is sent to a server, ever.

Works offline, always

Download a speech model once. After that, dictate anywhere, plane, basement, no Wi-Fi needed.

No account. No tracking.

No email, no sign-up, no analytics on your usage. Download and go.

No audio files stored

Recordings are transcribed and discarded instantly. Only text history stays, and only if you want it.

Open weights model

Powered by Whisper.cpp, open-source, auditable, community-trusted. No black box.

GPU-accelerated on Mac

Uses Apple Metal for near-instant transcription. Fast without a cloud subscription.

The overlay

Meet the Dictation Overlay.

A minimal floating pill that sits on your screen, shows live audio waves while you speak, and disappears into the background when you don’t need it.

WhisperVoca

Click to preview each state

Always visible

Stays on top of all apps and windows. Never disappears when you switch.

Drag anywhere

Place it where it feels right. Position is remembered between sessions.

Live audio waves

Animated bars pulse while you speak, you always know you're being heard.

Right-click menu

Quick access to settings, microphone picker, hide options, and quit.

Customise it

Change pill colour, audio indicator colour, and live preview before saving.

Hide when needed

Show, hide, or hide for 1 hour. Reset position if it drifts off screen.

Speech models

Choose your speed. Choose your accuracy.

Download the model that fits your workflow. Switch anytime. Everything runs on your machine.

Tiny

Fastest

Quick notes, fast typists, lower-end machines

Speed95%
Accuracy70%
Model size75 MB
Great default

Base

Recommended

Everyday dictation, the recommended starting point

Speed80%
Accuracy84%
Model size145 MB

Small

Most accurate

Accents, technical terms, and faster speakers

Speed60%
Accuracy94%
Model size460 MB

Download directly inside the app, with a progress bar. Delete anytime to free up space.

Transcription history

Never lose what you said.

Every dictation is saved locally, grouped by day, and timestamped. Copy any past entry in one click. New transcriptions appear as you dictate, no refresh needed.

  • Grouped by Today, Yesterday, and earlier dates
  • One-click copy of any past dictation
  • Updates live as you dictate
  • Stored only on your machine
Transcription History
Today
6:42 PM

Schedule the team sync for Thursday at 2pm.

4:11 PM

Follow up with Sarah about the design review, she mentioned she needs the assets by end of week.

2:33 PM

Add error handling to the upload endpoint and write a test for the timeout case.

Yesterday
5:20 PM

The onboarding flow needs a loading state between the model download and first use.

11:08 AM

Reply to Jake: sounds good for Monday, can we do 10am instead?

Built for real use

Everything you need.

Core

Hold to record, release to paste

The most natural dictation flow possible. Hold your global hotkey anywhere, speak, let go, text appears instantly in whatever field is active.

Privacy

100% offline & private

Your voice is processed locally by Whisper AI. Nothing leaves your machine. No cloud, no server, no subscription, ever.

Smart

Hallucination filter

Silence and empty recordings are quietly discarded. Whisper won't paste garbage like "Thank you for watching" into your document.

Works in any app

OS-level paste, if it has a text field, WhisperVoca works there.

8 dictation modes

General, Email, Slack, Code, Dictation, Medical, Legal, Finance. Pick the one that fits what you're writing.

Filler word removal

Strips "um," "uh," "you know" automatically. Polished output by default.

Transcription history

Local log grouped by day. Copy any past dictation in one click. Updates live.

Custom vocabulary

Teach WhisperVoca your names, brands, and jargon. Auto-learn from edits in history.

Multiple speech models

Tiny → Small. Download inside the app, switch anytime, delete to free space.

Customizable overlay

Change the pill color, wave color, drag it anywhere on screen. Hide it for an hour if you need the space.

Mic device picker

Choose built-in mic, headset, USB mic, or follow system default.

No subscription

Pay once, own it forever. No monthly fees, no feature paywalls, no account.

Mac & Windows

Same experience on both. One purchase, both platforms.

Native right-click menu

Quick access to settings, mic picker, hide options, and quit from the overlay.

GPU-accelerated on Mac

Apple Metal under the hood. Near-instant transcription without a cloud subscription.

What’s next

The roadmap.

WhisperVoca is actively developed. Here’s what’s coming next.

Custom hotkeys

Remap hold-to-record to any key combination you like.

Auto-stop on silence

Stop talking and transcription kicks off on its own. No key to release.

Medium model

Higher accuracy for complex vocabulary and technical language.

Launch at login

WhisperVoca is ready the moment you sit down. No manual start.

Menu bar / system tray

Run it in the background without the floating overlay.

Custom speech models

Bring your own Whisper-compatible models and install them straight in the app.

Ready to stop typing?

One-time purchase. No subscription, no cloud, no ongoing fees. Download and you're dictating in minutes.

100% local  ·  No account required  ·  Mac & Windows  ·  Powered by Whisper AI