Hold a key. Speak. It's typed.
Keebye is push-to-talk voice dictation for macOS, built for developers running several AI coding agents at once. Hold Right ⌘, say the thing, release — and the text lands in whichever app has focus: Claude Code, Cursor, ChatGPT, Slack, or a terminal session over SSH inside tmux.
Tap the hotkey once instead of holding for longer dictations, press Esc to cancel, and remap to Fn or Right ⌥ if Right ⌘ isn't your key.
Built for the terminal
Most dictation tools fall apart the moment you leave a text box. Keebye pastes with terminal-aware chunking, and an optional synthetic typing mode survives SSH sessions, tmux panes and non-QWERTY layouts.
Say it messy, it lands clean
- Fillers, false starts and doubled words are stripped by deterministic rules
- An optional polish pass runs a local LLM, also on-device
- Turn cleanup off and you get every word verbatim — no model deciding what you meant
- A custom dictionary teaches Keebye your product names and acronyms once
On-device, by design
Audio never leaves your Mac. The network is used only to download models, check for updates and confirm your plan — never to send audio, and there's no telemetry or analytics. Once a model is downloaded, dictation works with the Wi-Fi off.
History is stored locally as text only and auto-deletes after 30 days. Keebye also refuses to type into secure password fields.
The default engine is English-tuned and fast; switch to the multilingual Canary engine for 25 languages on-device. Free 14-day trial, no card required.
