PENGEMBANGAN APLIKASI DETEKSI DARURAT BERBASIS PENGENALAN SUARA MENGGUNAKAN METODE MOBILE APPLICATION DEVELOPMENT LIFE CYCLE

Sexual violence remains a serious social issue in Indonesia and can happen
to anyone, especially women. This situation calls for a technological innovation
that enables victims to seek help quickly without needing direct physical interaction
with their devices. This study aims to develop an Android-based emergency
detection application using speech recognition capable of identifying specific
emergency keywords as danger signals and automatically sending messages and
real-time location data to emergency contacts. The development process follows
the Mobile Application Development Life Cycle (MADLC) method, consisting of
the identification, design, development, prototyping, testing, and maintenance
phases.The application was built using the Kotlin programming language on the
Android platform and integrates Firebase Realtime Database, Google Speech API,
and Fused Location Provider services. Evaluation was conducted through black-
box testing to verify functionality and through accuracy experiments based on
sound intensity (dB) under various environmental conditions. The results show that
the system successfully recognizes the keyword at sound intensities of ≥61 dB with
100% accuracy at a distance of one meter, while accuracy decreases to 60–70% in
noisy environments such as busy roads (~78 dB).These findings indicate that the
effective detection threshold lies around ≥61 dB, and the application has strong
potential as a practical early-warning tool for preventing and responding to sexual-
violence emergencies.