Credit: Burst from Pexels

Post-traumatic stress disorder (PTSD) is a mental health disorder triggered by traumatic experiences such as accidents, disasters, or violence. Exposure therapy, a scientifically validated treatment for PTSD, involves the gradual re-exposure of patients to trauma-related cues to reduce anxiety and avoidance behaviors.

One major challenge in exposure therapy is tailoring recreated environments to individual patients, a process that is often time-intensive. Although virtual reality (VR)-based approaches have been explored in recent years, patient-specific customization remains difficult owing to the effort required to design bespoke environments.

In a new study published in ACM Transactions on Computing for Healthcare, researchers from the University of Tsukuba developed a system that automatically generates auditory VR experiences of traumatic sounds from natural language input by leveraging large language models and acoustic datasets. Users simply input a theme in text, and the system automatically generates appropriate sound materials and scenarios.

Clinical experts who tested the system reported that it was straightforward to use, produced realistic sound quality, and could be implemented in therapeutic settings with only minor adjustments. Experimental trials showed that generating a single auditory VR session required an average of five minutes (with a minimum of 50 seconds), meeting the time constraints of clinical practice.

Looking ahead, the researchers aim to improve the precision of the system in handling complex auditory environments with multiple overlapping sounds and to pursue further clinical validation for practical deployment.

More information: Yuta Yamauchi et al, Development and Evaluation of an Auditory VR Generative System via Natural Language Interaction to Aid Exposure Therapy for PTSD Patients, ACM Transactions on Computing for Healthcare (2025). DOI: 10.1145/3723048