​
Input

  • Required Inputs:
    • actual_output: URL of the audio recording to evaluate

​
Output

  • Result: Count of inappropriate interruptions (integer)
  • Reasoning: Brief explanation of detected interruptions

​
Interpretation

  • Higher counts: More frequent user interruptions
  • Lower counts: Smoother responses and better engagement