Home / Categories / Innovation List / Innovation Details
Image

Quiet Zone - App prototype for noisy environment detection system with Machine Learning

Description: Android sound meter app prototype for loud environment detection system with a built-in Machine Learning Classification algorithm.

Recently, on the internet, many videos with younger and younger people that have complained about hearing problems have surfaced on the internet, indicating the fact that this issue is appearing at lower and lower ages. The main concern is that the world’s population does not take this issue seriously. We are trying to raise awareness so that humanity can preserve its hearing. Especially in our age, an era in which the streets of the cities are flooded with the deafening honking of cars and intense use of working machinery. With our technology, we can produce a tremendous impact in the quality of life of human kind. Our first prototype was based on machine learning AI and Arduino IDE board. The purpose of the Arduino board was to record the noise levels surrounding it using a microphone while storing them in a text file, but to also prove that our technology is viable on any device which has a microphone. The AI uses the Random Forest Classification method, which has 3 principles: Decision trees, Ensembled learning, and Bootstrapping. We first trained the AI with a set of dummy values with over 390 rows of data with regard to the open-source data of NIOSH, thus resulting in a 94% accuracy rate. This Machine Learning model is able to provide an efficient diagnostic depending on the user’s age, the time exposure to the noisy environment, and the sound level. For the model, we made an Android implementation of the app called “Quiet Zone”. Regarding the front-end, we used Flutter to create a user-friendly interface that is easy to use by any age group. You start by entering your age, and then giving the app permission for the microphone and to send notifications. Then, at the press of a button, the app starts storing the noise level values, which will be used by the AI. The back-end has a modified code to access your phone’s microphone and an API code that connects the app to a 24/7 open server, containing the AI code. The app is compatible with all Android phones. In order to develop our app, there are some additions and changes that we can make. The first one is the iOS and smartwatch implementation so it can be used by a larger audience. This addition would not be difficult to make because we used Flutter. Another one is optimization of the app, regarding the resources it takes and also the interface’s performance. Besides these 2, we are looking forward to implementing the feature of noise level tracking, making it easier for them to find patterns in their lives. To conclude, our product represents a safe and natural way in which the users can protect their hearing without resorting to hearing aids that would cost them a fortune. In addition, this would also raise awareness about the problem, which can positively improve their overall life because many people do not know how grave and frequent this issue is around the globe.

Organisation: Colegiul Național Nichita Stănescu Ploiești

Innovator(s): Constantino-Daniel Boșcu, Mihai Alexandru Bizineche, Andrei-Daniel Boghici, Alexandru Dornea, Dominic Luca Ioan

Category: Young inventors

Country: Romania

Gold Award