Desktop · Voice-to-Text

Talk.It types.

Murmur is a private, on-device dictation app for developers. Press a hotkey, speak naturally, and your words are typed straight into any app, whether editors, terminals, or chat. Nothing leaves your machine.

Download for Windows View on GitHub

Detected Windows · free, signed, and auto-updating

  • 100%On-device
  • 5STT models
  • 0Cloud calls
  • v0.4.1Latest build
// What makes it different

Everything local.
Everything fast.

01

Private by default

Every byte of audio is transcribed on your own machine. No cloud, no telemetry, no network calls in normal use.

0 network calls
02

Live preview

Your words appear as you speak. A caption follows your active window, macOS-style, or sits under the floating pill.

See it before it lands
03

Types into anything

Direct keystrokes land in editors, terminals, and chat. Output never routes through the clipboard, so nothing leaks.

Editors · terminals · chat
04

Voice editing

Editing commands like "select all", "copy that", "undo that", "scratch that", and "new line", recognized inline as you talk.

Hands-free editing
05

Multilingual + translate

Transcribe in many languages, or translate your speech straight to English with the multilingual model.

99+ languages
06

Fast and local

whisper.cpp and NVIDIA Parakeet with optional CUDA. Pick base, small, medium, large-v3-turbo, or Parakeet.

GPU-accelerated

How it works

Audio never leaves the device. The whole pipeline runs locally, from microphone to keystrokes.

  • 1
    MicrophoneCPAL · 48 kHz
  • 2
    Silero VADspeech gate
  • 3
    On-device STTParakeet · whisper.cpp
  • 4
    Keystrokesinto any app

Built with

RustTauri v2whisper.cpp (whisper-rs)ONNX RuntimeSilero VADNVIDIA ParakeetCPALCUDAenigoglobal-hotkey

Get Murmur

Free and open source. The Windows installer is signed and updates itself when a new version ships. macOS support is in progress.

  • Windows 10 or 11 (x64)
  • An AVX2-capable CPU (any Intel or AMD chip from about 2013 on)
  • An NVIDIA GPU is optional and accelerates the larger models
  • About 150 MB, plus whichever model you download
Download v0.4.1 Source on GitHub Detected Windows · free, signed, and auto-updating