Dictus Desktop
Offline voice dictation transcribes speech into any app on macOS, Windows and Linux using Whisper or Parakeet. Free, MIT licensed and account-free.
Dictus Desktop is a free, open source voice dictation app for macOS, Windows and Linux. It turns speech into text in the app you're typing in, with transcription handled entirely on your computer. It's for people who want to dictate messages, documents or emails without sending their recordings to a cloud service.
You don't need an account. A global shortcut starts and stops recording, and push-to-talk is also available. Dictus inserts the result into the focused app, so you can use it across your existing writing tools rather than copy text out of a separate transcription window. Voice activity detection filters silence during recording.
You can choose Whisper or Parakeet V3 for on-device speech recognition. Whisper supports CPU processing and GPU acceleration, with a GPU recommended for its larger models. Parakeet V3 runs on the CPU and detects the spoken language automatically. Its stated hardware requirement is an Intel Skylake processor, an AMD equivalent or newer. The macOS app supports both Apple Silicon and Intel Macs; Whisper can use Intel, AMD or NVIDIA GPUs on Windows and Linux.
Dictus uses the MIT license, so its code is available to inspect and modify. Local transcription works without an internet connection, and the app collects no telemetry or analytics. Linux builds are available for x64 and ARM64, with packages for Ubuntu, Debian, Fedora and openSUSE as well as an AppImage.
