Voice-to-text,that understands
Hold Fn to talk, or double-tap Fn to toggle. Your words are transcribed, cleaned up in context, and pasted into the active app — all in under a second.
Universal binary — Apple Silicon + Intel · Signed & notarized · MIT licensed
Dictation that
gets you.
Voxa Voice reads your screen, understands your context, and types for you. Built for people who think out loud and want their Mac to keep up.
Context-aware transcription
Voxa reads the frontmost app, window title, and selected text to spell names correctly and match tone. Replying to an email? It knows who. Dictating into a terminal? It formats like a command.
Download.Key.Talk.
Your keys,
your rules.
Two defaults out of the box, infinite combinations if you want. Hold, toggle, or chain them. Every shortcut is rebindable from Settings.
Press and hold the globe/Fn key. Voxa records while the key is down, transcribes on release, and pastes straight into the active field.
Start in hold mode, then double-tap to latch into toggle mode — without stopping the recording.
Just say
“Hey Voxa.”
Skip the keys. Say the wake word and Voxa wakes up, reads your active window, and does what you asked — using the full context of whatever you’re looking at.
Always listening for the trigger when the app is running. Nothing is sent anywhere until Voxa hears its name.
Voxa reads the active window via macOS Accessibility — only after you call it by name.
Under-a-second
pipeline.
Every app,
every text field.
Voxa Voice types via the macOS Accessibility API. If you can type into it, you can dictate into it — with context-aware cleanup tuned to whatever you have open.
Your voice,
your data.
Voxa Voice was built so you never have to think about where your words go. No accounts, no history, no analytics — your voice gets turned into text, and that's the end of the story.
No accounts, no sign-up
No email, no login, no telemetry. The app never asks who you are — it just turns your voice into text and gets out of the way.
Audio stays ephemeral
Recordings live only long enough to be transcribed, then they're discarded. Nothing is saved to disk, nothing is queued for analytics.
Zero retention
Voxa keeps no server logs because Voxa has no server. Transcripts exist only in the app you pasted them into — nowhere else.
Permissions you can see
Only microphone and accessibility access — granted by you, revocable at any time in System Settings. Nothing hidden, nothing sneaky.
Native Swift.
Feels like macOS.
A real Mac app — universal binary, menu-bar native, notch-aware. MIT licensed, built for people who live on their keyboard.
Native Swift
Built directly on AVAudioEngine, ScreenCaptureKit, and SwiftUI. No Electron, no web shell, no battery tax.
Menu bar only
Lives in your menu bar. No dock icon, no window clutter. It's there when you hold the key, invisible when you don't.
Notch-aware overlay
The recording indicator hugs the notch on newer Macs and tucks into the menu bar on older ones. Designed for macOS, not ported to it.
MIT licensed
Use it at work, bundle it with your team, ship it in your stack. No strings attached.
Loved by early adopters.
Fn-to-talk is unreasonably fast. I hold it, I speak, and my words just appear in the right place. It's the feature my Mac has been missing for years.
Arjun Mehta
Senior Developer, Early user
Stop paying for
dictation.
Download the .dmg, drag it into Applications, and hold Fn. That’s the whole setup. No account, no subscription, no friction.
macOS 13+ · Universal binary · MIT License
