Skip to main content

Voice Mode with CALM AI Tools

Voice Mode enables hands-free interaction with CALM Copilot Chat, allowing you to speak architecture queries and receive spoken responses directly in the chat interface.

As of this Voice Mode is supported only in VSCode with Github Copilot. The remainder of this document assumes this setup.

Getting Started​

Prerequisites​

  • VS Code: Version 1.94 or later
  • CALM Copilot Chatmode: Configured in your repository (see CALM AI Tools)
  • GitHub Copilot Chat: Active subscription and extension installed
  • Microphone: A working microphone

Setup​

  1. Install the VS Code Speech Extension

    • Open VS Code Extensions (Ctrl+Shift+X or Cmd+Shift+X)
    • Search for "VS Code Speech"
    • Click Install on the Microsoft-published extension

  2. Grant Microphone Permissions

    • First use will prompt for microphone access - click Allow
    • If prompted on macOS, grant permissions to both "Visual Studio Code" and "Code Helper", then restart VS Code
    • On Windows, ensure VS Code is enabled in Settings → Privacy → Microphone

  3. Configure VS Code Settings

    • Open VS Code Settings (Ctrl+, or Cmd+,)
    • Search for accessibility.voice and add these settings:
{
  "accessibility.voice.autoSynthesize": "on",
  "accessibility.voice.ignoreCodeBlocks": true,
  "accessibility.voice.speechTimeout": 1200,
  "accessibility.voice.keywordActivation": "chatInView"
}

Using Voice Mode​

Starting a Voice Query​

  1. Open GitHub Copilot Chat (Ctrl+Shift+I or Cmd+Shift+I)
  2. Select the CALM chatmode
  3. Click the microphone icon in the chat input field or press Ctrl+I (Windows/Linux) / Cmd+I (macOS)
  4. Speak your query clearly
  5. The response will appear as you finish speaking (auto-submit after ~1.2 seconds of silence)
  6. Your response will be read aloud automatically

Quick Voice Activation​

Say "Hey Code" to activate voice input without clicking anything (requires keywordActivation setting enabled above).

Example Queries​

Speak naturally and clearly. Here are examples of how you can use CALM Chatmode via voice:

Building Your Architecture:

  • "Create a microservices architecture with a web frontend, API gateway, and PostgreSQL database"
  • "Add a node for the authentication service with HTTPS interfaces"
  • "Create a connects-to relationship from the API gateway to the backend services"
  • "Define that the services are deployed in Kubernetes"

Defining Controls and Compliance:

  • "Add PCI-DSS compliance controls to the payment processor service"
  • "Define encryption-in-transit controls for all API connections"
  • "What security controls should apply to our database?"

Modeling Business Processes:

  • "Create a flow for the customer onboarding process"
  • "Model the trade execution flow from order placement to settlement"
  • "Show the data journey through our system"

Validation and Best Practices:

  • "Is my architecture valid?"
  • "What patterns apply to this microservices design?"
  • "What interfaces should the authentication service expose?"
  • "Why is my relationship definition failing validation?"

Documentation and Visualization:

  • "Generate documentation for my architecture"
  • "Create a visual diagram of the system"
  • "Document all the interfaces and their protocols"

Listening to Responses​

Responses will be read aloud automatically (if autoSynthesize is enabled). You can also:

  • Stop playback: Press ESC or click the speaker icon
  • Manual read-aloud: Click the speaker icon next to any response to hear it read aloud again