Hangarvibe

Product guide

HangarVoice guide

Speak your tasks instead of typing them.

What it is

HangarVoice lets you talk instead of type. You speak, and your words are placed into the terminal as text. You read them, fix anything wrong, then press Enter to send. The default engine runs on your own machine with Whisper, so it is private and needs no API key.

Before you start

  • Hangarvibe and a working microphone.
  • On Windows, the system microphone privacy setting must be turned on.

How to use it

  1. 1

    Click the mic on a bay, or on the assistant composer.

  2. 2

    Talk. Say what you want the agent to do.

  3. 3

    Read the transcript that appears in the terminal.

  4. 4

    Fix anything that is wrong, then press Enter to send it.

  5. 5

    Optional: to type into any app, turn on global dictation in Settings → Voice.

  6. 6

    Then press Ctrl+Shift+D to start dictation, and press it again to stop.

Good to know

  • The first time you use Whisper, it downloads a model once. It uses your GPU when you have one.
  • You can switch to a Groq or OpenAI cloud engine with your own key under Settings → Voice.
  • Add custom vocabulary rules to fix product names or casing in your transcripts.
  • Your words are only staged. Nothing is sent until you press Enter.

Related