Speech-based Psychological Crisis Assessment using LLMs

ArXi:2605.10027v1 Announce Type: cross Psychological hotlines provide critical for individuals experiencing mental health emergencies, yet current assessments largely rely on human operators whose judgments may vary with professional experience and are constrained by limited staffing resources. This paper proposes a large language model (LLM)-based framework for automated crisis level classification, a key indicator that s many downstream tasks and improves the overall quality of hotline services. To better capture emotional signals in spoken conversations, we.