You talk. It types.
An efficiency essential for vibe coders.
A voice-input app for macOS. Hit a hotkey, speak, and the text
drops right into the field you're in.
Four modes: raw transcription,
AI cleanup, translation, or fire up a terminal Agent to get real
work done.
Hotkeys, vocabulary, models, and terminal are all
yours to set up.
Tap to start, tap again to stop. Defaults are below — remap any of them.
Local ASR + filler removal + zh/en fixup. Text drops straight into the field.
De-um, rewrite, or draft an email. Select text first to act on the selection.
Speak once, land it in the language you set.
Hand your dictation, selected text, or Finder files straight to Claude Code, Codex, or any terminal agent.
Real captures, played at 1.8×.
SpellKey can be dead simple or deeply customizable. Great defaults out of the box; the details are up to you.
All four modes are remappable — and you can switch mode mid-recording.
Add names, jargon, and often-misheard words to your glossary; they double as hotwords for the engine.
Three speech engines. A local model handles cleanup, or plug in your own API.
You decide the agent, terminal app, launch command, and working directory.
Buy once, use it forever, updates always free.
✓14-day, no-questions-asked refund. Not for you? Get your money back. Refund policy →