Skip to content
buildbay.
Handy logo

Handy

cjpais · Audio

GitHub stars

32,500
—Collecting 30d data

Stats updated

32,293 stars on Sep 27 to 32,500 stars on Oct 1, up 207.5 measured star snapshots

What is Handy?

Handy keeps the dictation loop compact: trigger recording from the keyboard, speak, and paste locally transcribed text into the application already in focus. Windows, macOS, and Linux builds offer model choices and shortcut settings; setup requires the relevant microphone and accessibility permissions.

Handy is a free, open-source desktop speech-to-text app that runs locally. A configurable shortcut records your speech and pastes the transcription into the text field of the app you are using.

Core capabilities

  • Record from configurable hold, toggle, or push-to-talk keyboard shortcuts
  • Transcribe locally with Whisper-family or CPU-oriented Parakeet models
  • Paste into the active app and control recording through CLI or Raycast actions

From shortcut to active field

Handy supports hold-and-release, toggle, and mode-specific shortcuts. After recording, it filters silence, runs the chosen local recognizer, and inserts the text back into the application that had focus, keeping capture and output attached to the current desktop task.

  • dictation
  • shortcuts
  • local
  • text input

Models and integration controls

Whisper models and CPU-oriented Parakeet provide different recognition paths, with available acceleration depending on the model and machine. CLI controls can operate a running instance, although the project labels them largely beta. A community Raycast extension adds another control surface; it is not part of the core desktop application.

  • Whisper
  • Parakeet
  • CLI
  • Raycast

Where it fits

Use cases

  1. 01

    Dictate into everyday desktop software

    Trigger Handy while a document, message, issue, or form is active, speak naturally, and let the locally produced transcript appear in the focused text field without switching to a separate editor.

  2. 02

    Keep voice data on the workstation

    Choose a supported Whisper or Parakeet model and perform voice activity detection and transcription on the computer, avoiding a required cloud transcription request for ordinary dictation.

  3. 03

    Connect dictation to an existing workflow

    Use the documented command-line controls or community Raycast extension to start and stop dictation from existing desktop workflows. The README labels CLI support as largely beta.