Description
Bullying Decoder turns everyday classroom audio, previously unused, into a continuous stream of structured behavioural data, making verbal bullying measurable at scale. Trained on hours of real classroom interactions, the system analyses tone, context, repetition, and emotional intensity to decode the sentiment of the class. The system maps both classroom locations and the school timetable, enabling CCTVs to accurately attribute behaviour to the correct grade, even in shared spaces like music rooms and labs. These conversations are anonymised and aggregated into dynamic data layers - tracking frequency, severity, patterns, and focus areas over time. Visualised through a dashboard, the system maps behavioural trends across grades, transforming scattered conversations into a unified intelligence system. By converting unstructured audio data into prioritised, decision-ready insights, it enables timely interventions, moving schools from reactive discipline to predictive care.
