
murmure
Fully local, private and cross platform Speech-to-Text with LLM Post-processing
The Lens
By Erik Loyd, SaaS CEO and former COO/CFO of an AWS Premier Partner.
Updated Nov 2026
Murmure turns your voice into typed text anywhere on your computer, fully offline. Press a shortcut, talk, and the words land in whatever app has focus: an email, a doc, a coding agent prompt. It runs a tuned NVIDIA Parakeet speech model on your CPU, handles 25 European languages, and is free under AGPL-3.0.
Nothing to host. Installers ship for Windows, macOS and Linux, with Windows builds code-signed through the SignPath Foundation. Audio is deleted right after transcription and history lives in RAM, last five items only. Optional Prompt Mode pipes text through ollama/ollama or any OpenAI-compatible server to fix grammar or translate.
Verdict: a strong free replacement for paid dictation apps, and the privacy story holds because nothing leaves the machine unless you point Prompt Mode at a cloud model. Solo: free, install it today. Small teams: free, and command-line config import makes shared dictionaries easy. Larger teams: free, but no admin console or support contract.
The catch: this is dictation, not transcription. Recordings cap at five minutes and it only listens to a microphone, so it will not replace Otter for meetings or Rev for audio files. It is also essentially a one-person project.
Free vs Self-Hosted vs Paid
fully freeFree: The full app on Windows, macOS and Linux under AGPL-3.0, including Prompt Mode, voice mode, custom dictionary, formatting rules, the phone-as-microphone feature and the local API. No account, no telemetry.
Self-hosted: Runs on your own computer, CPU only, no GPU needed. Prompt Mode is optional and needs an LLM: Ollama locally (free) or any OpenAI-compatible server, local or cloud, where you pay that provider.
Paid: None. The developer takes tips through Tipeee; there is no paid tier.
Completely free. The only possible bill is a cloud LLM if you choose one for Prompt Mode instead of a local model.
What to do by team size
- Solo
- free, install and go
- Small team
- free, share settings with config import
- Medium team
- free, but no central admin or support
- Large team
- free for individuals; no fleet management or support contract
Get tools like this every Wednesday
One featured tool, three on the radar. No fluff.
Similar Tools
A low score is not a verdict on quality. Young and niche tools start low by design. How we calculate scores
Trust Signals
License: GNU Affero General Public License v3.0
Free to use; modified versions you distribute or offer over a network must be open sourced. The bundled speech model is CC BY 4.0.
Commercial use: ✓ Yes
About
- Owner
- 🇫🇷 Al1x-ai (User)
- Stars
- 1,166
- Forks
- 106
Explore Further
More tools in the directory
VoiceStudio
The open-source ElevenLabs alternative for local voice cloning, design, create, dubbing and dictation Desktop App
111.6k ★tldr
Collaborative cheatsheets for console commands 📚.
64.2k ★1Panel
🔥 1Panel is a modern, open-source VPS control panel, and the only one with native AI agent support. Run Ollama models, deploy OpenClaw agents, and manage your entire server stack from one clean web interface.
37.5k ★