Tap right ⌘, talk, tap again. Your words land wherever your cursor is, near instantly, because everything runs on your Mac. No account, no cloud, no minutes, free forever.
You need a Mac with Apple Silicon (M1 or newer). That's it.
1. Install Homebrew if you don't have it. Paste this in Terminal:
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
2. Install Yap. Paste this in Terminal:
git clone https://github.com/dertuman/yap && cd yap && ./install.sh
This downloads the speech model (574 MB, one time), builds the app, and puts it in Applications. Takes a few minutes.
3. Allow the two permissions when macOS asks: Microphone and Accessibility. If you missed the prompts, go to System Settings > Privacy & Security and turn Yap on in both lists.
4. Tap right ⌘ and start talking. Tap again to stop. Done.
- Tap right ⌘ to start, yap, tap again to stop. The text is at your cursor before you can look for it.
- Or hold right ⌘ and speak, release to transcribe, walkie-talkie style.
- Press any other key while holding and it cancels, so real shortcuts still work.
- The menu bar mic turns red while recording.
Want Yap to start when your Mac starts? System Settings > General > Login Items, add Yap.
I was using a paid dictation app and ran out of minutes. So I built this: the same thing with most features stripped away. It runs Whisper large-v3-turbo locally through whisper.cpp.
Five small Swift files. A menu bar app watches the right ⌘ key with an event tap, records 16 kHz audio, sends it to a local whisper-server (kept running so the model stays loaded and transcription is basically instant), then pastes the result at your cursor and restores your clipboard. Nothing ever leaves your Mac.
Want a different model? Grab any ggml model from Hugging Face, put it in ~/Library/Application Support/Yap/models/, and change the filename in Sources/Transcriber.swift.
MIT