Voice to text
Turn speech into text in any Mac app, not just the ones with a microphone button.
Vaitly Voice (VV) is a voice to text app for Apple Silicon Macs. Hold a shortcut, say what you mean, let go, and the words are pasted wherever the cursor is: Mail, Slack, Cursor, Notes, a browser. Recognition runs on the Mac, filler is dropped, and there is nothing to install inside each app.
Last updated
What a voice to text app actually does
Voice to text, dictation, speech to text: three names for one job, turning what you say into typed words. The difference between tools is where the words go and how much cleaning up you do afterwards. A note-taking app with a microphone button only helps inside that app. Built-in macOS dictation works in more places but writes down every um. VV is a system-wide voice to text app: it types into the frontmost app, whatever that is, and tidies the speech on the way.
Hold, speak, paste
Put the cursor where the text should go. Hold the shortcut you chose, speak as you think, and let go. VV recognises the speech on the Mac, using NVIDIA Parakeet for English by default, and pastes the result at the cursor. Your previous clipboard is restored once the app has taken the text. If both hands are busy, double-tap the shortcut instead to latch it hands-free, and tap again to stop. A small overlay shows that the microphone is live so you are never guessing.
Here is the kind of difference it makes. Said aloud:
“Um, so, hi Priya, um, can we move the, uh, the Thursday call to Friday morning, thanks.”
Pasted into Mail:
“Hi Priya, can we move the Thursday call to Friday morning? Thanks.”
The filler is gone and the punctuation is in. Nothing was rewritten; what you meant is what landed.
Works wherever there is a cursor
Because VV pastes through macOS Accessibility rather than a per-app integration, there is no plugin to add and no list of supported apps to check. Mail, Slack, Notes, Cursor, Xcode, Pages, a Google Doc in Safari, a form in Chrome, the Terminal: if you could have typed there, VV can type there. The speak into any Mac app guide covers the handful of apps that want a setting changed.
Your words, your names, your boilerplate
Two features make voice to text usable for real work rather than a demo. Vocabulary lets you tell VV how a name, product or acronym is spoken, so “Vaitly” comes out as Vaitly and not a guess. Snippets map a short spoken trigger to a block of text: say your trigger phrase and a full sign-off, address or reply template lands in one go. Both live on the Mac and work offline.
Limits and the privacy boundary
VV is transcription, not voice control. It will not select text, click buttons or act on “delete that”; it writes what you said. Accuracy depends on the microphone and on speaking at a steady pace, and English is the default model, with other languages available as optional downloads. On the privacy side, standard dictation audio never leaves the Mac. History and audio are stored locally, with a retention period you set in Settings. The only text that can leave is what you send to a cloud transform you configured yourself, and that is labelled in the app. See privacy for the full list.
Compared with the dictation built into macOS
Apple’s built-in dictation is free and already on your Mac, and for one sentence in a pinch it is fine. It keeps the ums and false starts, has no vocabulary or snippets of its own, and offers no history you can play back. VV vs Apple Dictation sets the two side by side. If you would rather see the mechanics first, start with how it works.
Questions
Which Mac apps does VV work in?
Any app with a text cursor. VV uses macOS Accessibility to paste where you would have typed, so Mail, Slack, Notes, Cursor, Pages, Safari and Chrome text fields all work without a plugin. If an app accepts pasted text, it accepts VV.
Is voice to text free on Mac with VV?
Yes, to start. The free plan gives you 1,000 words a week with no sign-in, and the count resets every Monday. Pro removes the weekly limit, and you can pay monthly, annually, or once for a lifetime licence.
Does my voice get sent to a server?
Not for standard dictation. Recognition runs on the Mac, so the audio is never uploaded. The only exception is a transform you have deliberately pointed at a cloud provider, which sends the transcribed text and is labelled as cloud in the app.
What Mac do I need?
An Apple Silicon Mac (M1 or later) running macOS 14 Sonoma or newer. Intel Macs are not supported because recognition relies on the Apple Silicon neural engine. The app asks for Microphone and Accessibility permissions on first run.
Can I dictate hands-free?
Yes. Double-tap the shortcut to latch it and VV keeps listening until you tap again. Holding the shortcut is the default because it is harder to leave on by accident, but latch is there for long stretches.