Abstract:
Deaf people face numerous challenges in their daily lives due to their inability to hear sounds in the environment. To address this issue, we have developed a smart visual alert system for deaf people that can detect and classify sounds in real-time and provide visual alerts using LED lights. The system is implemented on an edge computing device (a Raspberry Pi) to take the processing closer to data gathering in order to ensure fast and efficient processing of sound data and classification. The input sound is pre-processed to generate spectrograms which are then classified using a Convolutional Neural Network into several categories, including "Baby Cry", "Doorbell", "Talking", etc. The LED lights are controlled using GPIO pins on the Raspberry Pi, to provide different patterns or colors to indicate different types of sounds. A mobile app is also developed to allow users to view the history of events, adjust configurations, and access other assistive features that include Reminder, Speech-to-text, etc. The system has the potential to improve the quality of life for deaf people by providing fast and reliable visual alerts for important sounds at their homes.
