Click any field. Hold fn. Speak.

Why the fn key

The fn key sits at the bottom-left of every Mac keyboard. It doesn't do much by default — it modifies other keys but has no standalone function most users care about. That makes it ideal: it's always within reach, never accidentally pressed mid-sentence, and feels physically distinct from any typing key.

The hold-to-speak model is deliberate. You press, you speak, you release. There's no toggle to forget, no active recording light to check. The key is held down or it isn't — and that binary maps directly to whether you're dictating or typing. After a few sessions it becomes muscle memory.

Works everywhere your cursor is

Most voice tools require you to interact with them first. You open the app, dictate into their interface, then copy the result to wherever you actually needed it. That's two apps, two contexts, a clipboard operation.

Resonant works differently. It injects text directly at the system level — into whatever text field has focus at the moment you hold fn. The result lands exactly where your cursor is, without the intermediate step.

If a Mac app accepts keyboard input, Resonant works in it. There's no integration to install, no plugin to configure, no app-specific support to wait for.

What “perfectly formatted” actually means

Raw speech-to-text produces transcriptions — faithful but rough. The words are there; the structure isn't. You get run-ons where you paused for thought, filler words mid-sentence, missing capitalization, inconsistent punctuation.

Resonant applies a formatting pass on top of the transcript before injecting the text:

• Filler removal. "um", "uh", "like", "you know" — stripped silently. The text reads as if you never said them.
• Punctuation inference. Sentence boundaries, question marks, and commas are inferred from cadence and phrasing — not just pauses.
• Capitalization. Proper nouns, sentence starts, and acronyms capitalized correctly without having to say "capital".
• Paragraph breaks. Longer dictations with natural topic shifts are broken into paragraphs. You don't have to say "new paragraph".

The practical effect: you speak at a normal conversational pace and get text that looks like you wrote it deliberately. Most dictations require no editing at all.

Everything on your Mac, nothing on a server

Resonant transcribes using Apple's Neural Engine — the same chip that handles Face ID and other on-device ML. Audio never leaves your machine. There's no account required, no network request during transcription, no audio stored anywhere.

This matters more than it sounds. When you dictate into a Slack message, you're often discussing internal work. When you dictate into a Cursor prompt, you may be describing proprietary code or architecture. When you dictate a clinical note, the content is protected health information. Cloud dictation tools transmit that audio to remote servers regardless of what the text contains.

With Resonant, the audio is processed where it was captured. The only thing that moves is the finished text — and only because you chose to send it.

What it replaces

For most people who switch to this workflow, the fn key replaces one of three habits:

• Slow typing of long text. Emails, Slack messages, AI prompts, meeting notes — anything where you're converting thought to text and length matters. Speaking is three to four times faster than typing for most people. The gap widens when the content is complex.
• Abbreviated prompting. Most people undersupply context in prompts because typing the full thought is tedious. Voice removes that constraint. What would take two minutes to type takes twenty seconds to say — so you say it all, and the model response is proportionally better.
• The separate dictation app. Tools that require opening a dedicated interface, recording, reviewing, and pasting. Resonant collapses that into hold-speak-release. The tool disappears into the gesture.

Frequently asked questions