A research team at Universidad Carlos III de Madrid (UC3M) has developed an AI-powered system capable of detecting signs of gender-based violence through paralinguistic features in the human voice, such as tone, rhythm, and intensity. The technology, based on advanced machine learning and adversarial neural network architectures, identifies biomarkers in speech that correlate with psychological stress or trauma, offering a non-invasive and privacy-preserving method for early detection. Published in the journal Applied Sciences, the study leverages 3D spectrograms of voice recordings to train the system to distinguish between individuals who have experienced gender-based violence and those who have not. The approach mimics how humans intuitively interpret emotional cues in speech, but with the added precision and scalability of artificial intelligence. The research was conducted using virtual reality environments where volunteers were exposed to videos with and without violent content. Researchers monitored changes in participants’ vocal patterns and emotional responses. The results revealed significant differences in how individuals with a history of violence reacted compared to those without such experiences—leading to a key discovery: voice characteristics alone could signal past victimization. “This was a serendipitous finding,” said Carmen Peláez Moreno, professor in the Department of Signal Theory and Communications at UC3M and researcher at UC3M4Safety. “While studying emotional responses, we realized that voice patterns could serve as reliable indicators of prior exposure to gender-based violence.” The technology holds substantial promise for real-world applications. It could be integrated into telephone helplines, telemedicine platforms, and virtual assistants to automatically flag potential cases of gender-based violence during calls or consultations. By identifying at-risk individuals early—often before they recognize their own trauma—the system could enable timely intervention, reduce underreporting, and support faster psychological recovery. “This technology could help prevent tragedies by enabling support before a crisis escalates,” Peláez added. “It allows us to act proactively, even when the individual hasn’t yet sought help.” The project is part of the broader Bindi initiative, led by the UC3M4Safety team, which aims to combat gender-based violence through prevention, evidence collection, and early intervention using technology. The multidisciplinary team includes experts from UC3M’s School of Engineering, the Institute for Gender Studies (IEG), and various academic departments across the university, drawing on more than fifteen fields, including engineering, psychology, social sciences, and humanities. Celia López Ongil, director of the IEG and professor at UC3M, emphasized the team’s mission: “We believe technology can play a transformative role in addressing social challenges. By combining innovation with a deep understanding of gender violence, we aim to empower victims, strengthen support systems, and contribute to a safer society.”