Building an Aroused Voice Detection Model to support self tracking of occurrences of Intimate Partner Violence from Noisy, Poor Quality Mobile Phone Recordings

Haiven is an organization that is working to use technology to improve the safety and wellbeing of individuals experiencing intimate partner violence. One of our projects is a mobile application, which helps to improve the safety of users. Through this project we hope to expand on our application by adding an audio event detection that, once recording is triggered, will tag individual sound. Specifically we are interested in identifying individual voices, shouts and screams. Once these events are tagged the user will be able to journal and assign sentiments to these individual audio events. Overall, our goal is to help victims gain a more thorough understanding of patterns in the abuse they experience and collect evidence to be shared with their counselors as well as the court system.

Paige Tuttosi
Faculty Supervisor: 
Parvaneh Saeedi;Anita Tino
British Columbia
Partner University: