Skip to main content

Voice Agent

Purpose​

The voice agent captures spoken commands and converts them into text to be sent to the VLA pipeline.
It is the first step in the Capstone workflow.

How it Works​

  1. Captures audio from a microphone.
  2. Converts audio to text using speech recognition.
  3. Sends text to the VLA pipeline or main agent.

Python Implementation​

  • examples/capstone/voice_agent.py contains the actual Python code.
  • Key functions:
    • listen_command() – listens to the microphone and returns a text command.
    • Handles errors like unrecognized speech or API issues.

Independent Test​

  • Speak a command into the microphone.
  • Verify that the recognized text matches your speech.
  • Ensure the text can be sent to the main agent for execution.