Skip to content
Get packs

Guide · Open source

How to use OpenWhispr: install, dictate and choose local or cloud models

Updated · written with AI from the project's own documentation and checked against our catalogue data

OpenWhispr is a free, open-source desktop app for dictation. You press a hotkey, speak, and your words are typed or pasted where your cursor is. This guide covers installing it, how dictation works, choosing between local and cloud speech models, and the limits to know before you rely on it.

OpenWhisprOpenWhispr/openwhispr · ★ 10.5k · MIT

Quick start

  1. 1Download or install. Get the installer for your platform, or run brew install --cask openwhispr on macOS or x86_64 Linux.
  2. 2Open the app. Launch OpenWhispr and let it finish first-time setup.
  3. 3Pick a speech model. Choose a local model such as Whisper or Parakeet for offline use, or add your own API key for a cloud provider.
  4. 4Set your hotkey. Choose the global hotkey you will press to start dictation.
  5. 5Dictate. Click into a text field, press the hotkey, speak, and check the pasted text.
  6. 6Try optional features. Explore the voice assistant, translation, meeting transcription or notes once dictation works.

Install OpenWhispr on macOS, Windows or Linux

Download the installer for your platform from the project's release page. macOS has a build for Apple Silicon and one for Intel. Windows uses an .exe file. Linux offers .AppImage, .deb, .rpm and .tar.gz files.

On macOS or x86_64 Linux you can also install the community-maintained Homebrew cask. It is maintained by the community rather than the project itself.

  • macOS (Apple Silicon): .dmg
  • macOS (Intel): .dmg, but live speaker identification and voice fingerprinting are unavailable
  • Windows: .exe
  • Linux: .AppImage, .deb, .rpm or .tar.gz
brew install --cask openwhispr

Dictate text into any app with a hotkey

Press the global hotkey, speak, and OpenWhispr types or pastes the transcribed text at your cursor. Because the hotkey is global, it works in any application that accepts text input.

OpenWhispr also has a dedicated dictation translation hotkey. You dictate in one language and the text is pasted in another.

  • Click into the text field where you want the words to go.
  • Press the dictation hotkey and speak.
  • Press it again or stop speaking, depending on your settings, and check the pasted text.

Choose local or cloud speech recognition

OpenWhispr can run speech recognition on your own device with local models such as Whisper, NVIDIA Parakeet, Orukeet and Cohere Transcribe. With local models, your audio never leaves your device.

If you want cloud processing for speed, you can bring your own API keys. The project lists OpenAI, Anthropic, Gemini and Groq among the providers you can use. Cloud features require your own keys.

Local Whisper can use GPU acceleration on Metal, CUDA and Vulkan (AMD and Intel).

  • Fully local: audio stays on your device; speed and quality depend on your hardware.
  • Cloud with your own key: faster on weaker machines, but audio is sent to the provider you choose.

Use the voice assistant and AI agent (optional)

A separate Voice Assistant hotkey sends what you say to your AI assistant as a command, with no wake word and no cleanup pass. Highlighted text can be edited in place. With auto-paste on, answers paste at a focused text cursor, or appear in a floating panel and are copied to the clipboard when no writable cursor is available.

You can opt in to sending a screenshot of your current screen as context.

Use meeting transcription and notes

Meeting transcription can auto-detect Zoom, Teams and FaceTime calls, with live speaker labelling. Notes let you create, organize and search notes in folders. The README also describes calendar integration with Google, Microsoft and Apple Calendar.

On Intel Macs, live speaker identification and voice fingerprinting are unavailable. Meetings still record and transcribe normally, and notes search falls back to keyword matching instead of semantic search.

Run OpenWhispr from source (for developers)

If you want to read or change the code, clone the repository and run the desktop app with npm. The project says this requires Node.js 24 or later. Full build details are in the project's documentation.

The mobile app is a separate Expo project in its own folder, with its own setup.

git clone https://github.com/OpenWhispr/openwhispr.git cd openwhispr npm install npm run dev

What people use it for

  • Dictating emails, documents or chat messages into any app without sending audio to a cloud service
  • Using your own OpenAI, Anthropic, Gemini or Groq key for cloud transcription
  • Transcribing meetings on Zoom, Teams or FaceTime, on Mac, Windows or Linux
  • Reading or modifying a desktop speech-to-text app's source code

Frequently asked questions

Is OpenWhispr free?

The project's README describes OpenWhispr as free and open source under the MIT licence, which allows free personal and commercial use. Cloud features require your own API keys from the provider, and those providers may charge you.

Does OpenWhispr work offline?

Yes, if you use local models. The README says local models such as Whisper and NVIDIA Parakeet keep your audio on your device. Cloud features need an internet connection and your own keys.

How do I install OpenWhispr on Mac?

Download the .dmg for Apple Silicon or Intel from the release page, or on macOS run brew install --cask openwhispr. The Homebrew cask is community-maintained.

Does OpenWhispr work on Linux?

Yes. The README lists .AppImage, .deb, .rpm and .tar.gz downloads for Linux, and a community Homebrew cask for x86_64 Linux.

Is OpenWhispr a good alternative to paid dictation tools?

It is a free, open-source option for dictation on Mac, Windows and Linux. Compared with paid tools, you handle setup yourself, and cloud features need your own API keys. Comparable projects include Whisper.cpp, Handy and Vibe.

Why are speaker identification features missing on my Intel Mac?

The README says live speaker identification and voice fingerprinting depend on ONNX Runtime, which stopped shipping macOS x86_64 binaries in version 1.24. Meetings still record and transcribe normally.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

OpenWhispr suits people who want system-wide dictation on Mac, Windows or Linux, with the option to keep audio on their device, and developers who want to read its code. It is less suited to people who want a zero-setup install with no API keys, or to Intel Mac users who need speaker identification. If you only need mobile dictation, the project's mobile app has separate setup.

Install options & downloads