Download OpenVoiceFlow

OpenVoiceFlow is a push-to-talk voice dictation app for macOS 14 or later, free and open source. One universal, signed, notarized DMG works on Apple Silicon and Intel Macs.

Universal native build

OpenVoiceFlow for macOS 14+

One universal DMG for Apple Silicon and Intel Macs.

macOS 14+ Signed, notarized, and stapled by Apple.
Download OpenVoiceFlow for Mac Install guide
  1. STEP 1

    Drag it to Applications

    Drop the app into the folder — that’s the install.

  2. STEP 2

    Open it from Applications

    Nothing is broken — open it from Applications, not the disk image.

  3. STEP 3

    macOS will warn you once

    “downloaded from the Internet” → click Open. You only see it once.

Native universal build

Apple Silicon and Intel · macOS 14 or later · 6.1 MB installer.

Universal macOSCURRENT

One signed, notarized build for every supported Mac

OpenVoiceFlow-0.5.29.dmg
sha256: c7a5d37996af03dfac8923c46c6308b784fb3d7791d8198ead76fb680f4ad2c8

On macOS 12–13? Use the retained OpenVoiceFlow 0.3.6 Apple Silicon fallback. The native 0.5.29 app requires macOS 14.

Download notes and limitations

  • The primary website-hosted DMG is the verified OpenVoiceFlow v0.5.29 release build.
  • OpenVoiceFlow v0.5.29 supports macOS 14+ on Apple Silicon and Intel. Windows and Linux downloads are not currently available.
  • iOS and Android apps are in the works — store links will appear here when they ship. Until then, no mobile version is available.
  • For macOS 12–13, the retained OpenVoiceFlow 0.3.6 Apple Silicon fallback remains available.

Download FAQ

How is OpenVoiceFlow licensed?

The source is public and use is free for everyone, including at work, under the GNU Affero General Public License v3.0. A modified version that is distributed or offered over a network must publish complete corresponding source under the same license, and the required attribution notice must be retained.

Does OpenVoiceFlow send audio to the cloud?

No. Audio transcription runs locally with WhisperKit by default, or with the separately downloaded Whistle (Beta) engine if selected in Dashboard Settings on Apple Silicon. Cleaned transcript text only goes to a cloud LLM if you turn on cloud cleanup.

Which backend should I choose?

Cleanup is Off by default (raw on-device transcript). Turn it on and pick OpenRouter — one key reaches any model — a direct provider (OpenAI, Anthropic, or Groq) with your own key, or local Ollama to keep cleanup fully on-device.