Local speech-to-text for macOS. Press a key, speak, text appears.
KeySpeak puts a waveform icon in your menu bar. Press your chosen shortcut anywhere on your Mac — in any app — speak, press it again, and your words are typed at the cursor.
Everything runs locally using Whisper. Your audio never leaves your machine.
- 🎙 Works everywhere — any app that accepts text input
- 🔒 Fully private — no internet, no API keys, no servers
- ⚡ Fast — Apple Silicon Metal acceleration via whisper.cpp
- ⌨️ Configurable hotkey — any modifier + key combination
- 👁 Live preview — see your words as you speak
- ✏️ Word corrections — fix names and jargon Whisper gets wrong
- 🌐 14 languages supported
- macOS 13 (Ventura) or later
- Apple Silicon (M1/M2/M3/M4) recommended
- ~500 MB disk space for the Whisper model
Option 1 — Buy the compiled app: Get the ready-to-use DMG at keyspeak.shop ($5, one-time).
Option 2 — Build from source (free):
# Install prerequisites
brew install cmake
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
# Download the Whisper model
mkdir -p ~/Library/Application\ Support/keyspeak
curl -L "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-small.bin" \
-o ~/Library/Application\ Support/keyspeak/ggml-small.bin
# Build and run
cargo run| Permission | Why |
|---|---|
| Microphone | To record your voice |
| Accessibility | To type transcribed text into other apps |
All audio is processed locally. No data leaves your machine. No telemetry, no analytics, no network requests.
MIT — see LICENSE. Purchase a compiled release at keyspeak.shop.