Silent Siren

Team Name

Silent Siren

Timeline

Summer 2020 – Fall 2020

Students

  • Alvaz Khan
  • Allea Cambell
  • Arpan Patel
  • Martha Taffa

Abstract

With the advancement in technology, a lot of work has been done for the welfare and betterment of people globally. Nevertheless, there are not as many systems implemented to inform deaf or hard of hearing (DHH) people when an emergency alert is going off, or when another such situation is taking place. This project is about the implementation of a mobile app that gives different pop-up alerts when detecting emergency sounds. The Silent Siren app aims to make a safer world for people by informing them about the critical situations taking place around them. It aims at creating a world where everyone has an equal chance of taking appropriate measures when it comes to emergencies. Silent Siren can also be used by other people in situations where they will not be able to hear sounds such as sirens or ambulance alerts in their surroundings. The notification system allows users to choose the notification type, which is basically vibration strength and flashing lights.

Background

Technology advancement is making a real difference in solving important problems. The focus has especially been given to minorities and underrepresented people, as well as people with disabilities. Features such as translators, subtitles and captioning for audio, smart home devices, text-to-speech, virtual assistants, and video services have been a game-changer in that respect. Nevertheless, one of the areas where there is not a lot of work is in emergency alerting systems. There is an emergency alert system in the United States that requires wireless cable systems, satellite, radio, and TV broadcasters to address people within a few minutes when there is a national emergency. Additionally, there are some sophisticated systems like Amber Alerts that inform people when an abduction occurs or severe weather alerts for Hurricanes or Tornadoes. These warning systems are critical for state and local officials to deliver significant emergency information to the impacted parts of the community.

Project Requirements

  • The system needs to be able to differentiate between smoke alarms, fire alarms, police sirens, and ambulance sirens.
  • The system needs to be able to ignore background noise that is not some sort of emergency siren so as not to give false alarms to the user.
  • The system should provide settings to toggle vibration strength.
  • The system needs to use as little memory as possible while running in the background.
  • The application shall be coded using Java and Python.
  • The system shall detect the siren in less than 60 seconds.
  • The application shall be for an Android platform.
  • The system should allow the user to choose the emergency signals to be notified on.
  • The system should have a dataset of common sirens that it compares any oncoming sirens against.
  • Adequate documentation of all the code shall be done to ensure maintenance.

System Overview

This software that is mainly for deaf and hard of hearing people will be an application that will give notifications when it identifies different emergency signals in the surroundings. It can get advanced by training a model to classify emergency sounds that give an accurate or close to an accurate result. This can be by eliminating the distractions of background noises such as music, rain, and wind. Dataset collection is also used for identifying the different sounds of emergency and non-emergency sounds.

Results

The Silent Siren app is convenient and effective because of the microphone usage of mobile phones to identify the different emergency alerts. The Silent Siren was successful since we were able to create a mobile application that correctly deciphers various different alarms and notifies the user. The app currently has basic functionality and there is further room for improvement as well as additional functionality. 

Future Work

The accuracy of the audio classifier of Silent Siren can be improved by training a model using a bigger dataset with a more varied emergency signal. The location-sharing feature will be helpful for future implementations such as distance approximation and localization of emergency signals approaching user from a distance. Additionally, more emergency signals can be added to the application and make the notification settings specific for each depending on the user preferences.

Project Files

Project Charter (link)

System Requirements Specification (link)

Architectural Design Specification (link)

Detailed Design Specification (link)

Poster (link)

Zipped Latex Folder (link)

Code Documentation (link)

Video (link)

Source Code (link)

References

N/A

mat4074