Resonant — Voice layer
Resonant isn't an app you switch to. It's a layer that sits beneath every application on your Mac. Press a key, speak, and clean text lands wherever your cursor is.
No windows to manage. No copy-pasting. No app switching. Just a hotkey and your voice — system-wide.
The problem
Every dictation tool on the market asks you to leave what you're doing. Open a window. Record. Wait. Copy the text. Paste it somewhere. Go back to what you were working on.
That workflow has five steps too many. The reason keyboards work is that they're invisible — they don't interrupt what you're doing. They're a layer.
Voice input should work the same way. Not as an application. As infrastructure.
Dictation app vs. voice layer
App
Open dictation app
Layer
Press one key
Copy transcription
Text lands in place
Paste into destination
Already there
Fix formatting
Clean on arrival
Switch back to work
Never left
Where it reaches
Resonant works wherever you can type — not just in supported apps. If there's a cursor blinking, Resonant can put text there.
Gmail, Outlook, Apple Mail
Compose a three-paragraph reply in 15 seconds.
Slack, iMessage, Teams, Discord
Respond to threads without context switching.
Notion, Google Docs, Obsidian
Draft meeting notes, specs, and memos at the speed of thought.
Cursor, VS Code, Terminal
Dictate prompts, comments, and commit messages inline.
Claude, ChatGPT, Copilot
Speak full context instead of abbreviating your prompt.
Spotlight, Arc, Chrome
Voice queries, search bars, URL bars — anywhere text goes.
How a layer works
One global hotkey activates Resonant from any app, any window, any text field. No dock icon. No menu bar distraction. It appears when you call it and disappears when you don't.
Your cursor stays where it was. Your window stays focused. The text arrives in the field you were already looking at. You never leave your workflow.
No persistent UI. No sidebar. No floating window. When you're not using voice, Resonant doesn't exist. When you need it, it's one keypress away.
What happens in between
“hey so the uh the main thing I want to flag is that the timeline for the Q2 launch is looking pretty tight, like we probably need to cut the analytics dashboard from scope or push the date by two weeks, I'd rather push the date honestly because the dashboard is the main thing customers asked for”
The Q2 launch timeline is tight. We have two options: cut the analytics dashboard from scope, or push the date by two weeks. I recommend pushing the date — the dashboard is the most requested feature from customers and worth protecting.
Processed entirely on your Mac. Your voice never touches a server.
Architecture
Cloud dictation routes your audio to a remote server, processes it there, and sends text back. That round-trip adds latency, creates a dependency on internet connectivity, and means everything you say — passwords mumbled near the mic, sensitive business context, personal notes — exists on infrastructure you don't control.
Resonant processes audio on your Mac's Apple Neural Engine. The entire pipeline — speech recognition, punctuation, text cleanup — runs locally. No internet required. No audio transmitted.
A layer should be as reliable as the hardware it runs on. Not as reliable as your WiFi.
Free. Local. System-wide.
One hotkey. Every text field. No subscription, no cloud, no app switching. Just voice — everywhere.
Download for MacDRequires macOS 14+ · Apple Silicon