My friend requires hearing aid on a daily basis but she cannot hear you when you call her name from behind since most hearing aids on the market is front biased. This is done because of the lack of directional information a hearing aid can provide, so by narrowing the range of voice amplified, a more focused information will be provided to the users. This will be a danger if a car is coming from behind the person cannot hear about it and this project is aimed to solve this particular problem.
What it does
It employs google cloud speech to text to identify if the sound is calling your name or not, if it is, and it is also coming from behind you, then it will ring a sound to notify you that you should check what's behind you.
How we built it
We used sound sensors with arduino to determine where the sound is coming from and a phone as voice input to check what is being said since the arduino microphones are of poor quality. The board is connected to a phone via bluetooth and the app connects to google cloud for voice recolonization to determine whether the sound is important or not. This is then send to arduino to play a note if it is important and also from behind.
Challenges we ran into
Connection between arduino and phone via bluetooth and the app that connects the board with google cloud.
Accomplishments that we're proud of
Built a decent model for prototype and demonstration purposes as well as the app development that allows voice recolonization.
What we learned
Arduino bluetooth, how to call google cloud in an app, linux and python.
What's next for NewSound
Next it should be able to tell apart if it's just background noise or it is a car horn that requires immediate attention like one's name.