Voice to text, for developers. Fast, native macOS dictation and read-aloud, with one-key on-device cleanup of what you say. 100% on your Mac — no cloud, no accounts, no API keys.
Open the .dmg, drag Murmur to Applications, and launch it. Updates install
with one click. Requires macOS 11+ on Apple Silicon (M1 or later).
- Dictate anywhere — hold Fn, speak, release. On-device Whisper types it at your cursor, in any app.
- Clean it up — hold Fn + Ctrl to run your words through a local LLM (fix grammar and punctuation, drop filler) before it's pasted.
- Read aloud — ⌘⇧R speaks the selected text (or the clipboard) in a high-quality on-device voice.
- Private by design — Whisper, the LLM, and speech all run on your Mac; your voice never leaves it. See PRIVACY.md.
git clone https://github.com/letsgetrusty/Murmur && cd Murmur
./scripts/setup.sh # toolchain, deps, signing cert, first build + model
./scripts/dev.sh # build, sign, launch — use this, NOT `tauri dev`cd src-tauri && cargo test # Rust tests
npm test # frontend tests (from repo root)Grant Accessibility on first run, then re-run dev.sh — it uses a stable
self-signed identity so the grant survives rebuilds (a bare tauri dev breaks
the Fn tap). Logs stream to ~/Library/Logs/murmur.log.
Deeper docs: architecture · signing & permissions · releasing.
MIT — see LICENSE. Third-party components and models are attributed in THIRD-PARTY-NOTICES.md (all permissive).