Survivors of domestic violence in Indonesia are often silenced by fear, financial dependence, and stigma—and when they are most at risk, they are least able to reach for a phone and ask for help. A case study from East Kutai found that 60% of women surveyed experienced verbal abuse and 40% experienced physical or sexual violence, with local case data showing dozens of reports of domestic violence each year. Existing panic button apps fail here because they all assume victims can act.
Suara Rumah is an Android app that passively listens for acoustic patterns of acute physical violence, such as screaming or breaking objects, without recording or understanding the content of the conversation. When it detects an escalating pattern, the app discreetly notifies pre-selected emergency contacts via WhatsApp/SMS with the victim's location, audio evidence, and concrete next steps for formal reporting, all without the victim touching their phone.
We built it with a native Kotlin Android app that performs audio feature extraction on-device, paired with a Python/FastAPI backend that runs a pre-trained Random Forest classifier. We intentionally designed this prototype to detect acute physical violence only, rather than claiming to be able to solve every form of domestic violence.
Log in or sign up for Devpost to join the conversation.