Auditory Brainwave Entrainment System using Time-Series Analysis of EEG Signal

This paper is focused on improving people’s general inability to quality sleep. We designed a feedback system that can process brainwaves in the form of time-series EEG data and convert them into RGB color-coded based topological map-like images. The system then uses CNN architecture based on 2D convolution with ReLU activation functions and dense layers for image and frequency classification to detect the EEG frequency and then based on the concept of brainwave entrainment, produce binaural tunes at required slower frequencies, to help to relax and slowly move the brain activity spectrum down to delta waves, which ranges around 0.5-4 Hz. The feedback signals create a gradual shift in the received frequency, thereby entraining the mind to tend towards the desired slower brainwaves. The testing was performed on Jetson Nano and the resulting accuracy tested for RGB-based frequency classification was over 80% which led to a successful binaural tone generation to create a slow auditory output wave, thereby eventually increasing the delta activity in the brain.