The Development of an Empathic Conversational Agent Designed for a Mental Health Helpline Service

<p dir="ltr">This PhD thesis develops a comprehensive framework for an empathic conversational agent (CA) tailored to mental health helpline services. It investigates how empathy can be recognised and expressed through voice, using co-design workshops with stakeholders and evaluation by users. A voice analysis algorithms was developed to classify empathy in both human and CA voices, highlighting challenges in detecting this complex, subjective construct. The findings show that empathy can be computationally detected and perceived by users, while emphasising the importance of diverse annotators, co-design principles, and careful implementation. This work provides practical and theoretical contributions to the design of empathic CAs in mental health care and open avenues for further research.</p>