This paper addresses the challenge faced by call center operators supporting family members of individuals with Substance Use Disorders (SUDs), particularly the difficulty of detecting emotional distress in callers during high-pressure interactions. We developed a lightweight, real-time Speech Emotion Recognition (SER) prototype for integration into call center workflows. The system applies speech segmentation and emotion classification to detect emotional states such as sadness, fear, anger, and disgust. We evaluated the prototype by benchmarking it with the CREMA-D dataset and with a small-scale user study. Our results indicate technical feasibility with moderate classification accuracy. The findings highlight the potential for AI-driven emotion detection to enhance emotional awareness and responsiveness in addiction helpline environments, helping operators to respond more effectively and ultimately improving the experience and outcomes for callers.
