I will provide a specific topic/scenario and include non-verbal information conveyed when Speaker A speaks.

Your task is to:

1. Generate five one-turn dialogues between Speaker A and Speaker B, and indicate the relationship between Speaker A and Speaker B. The dialogue must meet the following requirements:
    
    - Speaker A's spoken text **must not explicitly reveal** the provided non-verbal information. However, you should insert the non-verbal information in parentheses at the appropriate position (e.g., 'I’m fine (coughing lightly)').
        
    - The spoken content of Speaker A should be meaningful and sufficient to test a speech AI model's ability to understand and respond appropriately.
        
    - Speaker B's response should consider **both** Speaker A's spoken content and the provided non-verbal information. The response should demonstrate **high emotional intelligence**, including appropriate word choice and vocal characteristics (e.g., tone, pace). This response will serve as the "golden label" for evaluating the emotional intelligence of the speech AI model.
        
2. Using the **same spoken content** from Speaker A, provide:
    
    - A response from Speaker B that does **not** consider any of Speaker A's non-verbal information.
        
    - A response from Speaker B based on a **different set of non-verbal information** (e.g., a different emotional state, paralinguistic cue, or environmental condition) for Speaker A.

## Output Requirements:

Please format your output in JSON with the following fields:
{
  "relationship": {
    "speaker_A": "role of Speaker A",
    "speaker_B": "role of Speaker B"
  },
  "dialogue_with_nonverbal": {
    "speaker_A": "spoken text from Speaker A with non-verbal information in parentheses",
    "speaker_B": "spoken text from Speaker B based on both spoken content and non-verbal information"
  },
  "response_without_nonverbal": "Speaker B's response ignoring non-verbal information",
  "response_with_alternative_nonverbal": {
    "alternative_context": "description of the alternative non-verbal or environmental condition for speaker A",
    "speaker_B": "spoken text from Speaker B based on the alternative non-verbal information"
  }
}


## Input
