Project: MindSwiper

MindSwiper: Determining sound direction from EEG

ECE 202 course project, Winter 2024
Teammates: Eeman Iqbal, Jean Serrano, Poorva Bedmutha, Rommani Mondal

The project’s goal was to play “directional” audio and determine the audio direction from EEG signals. Direction was imbued to white noise by introducing a delay between the left and right channels in an attempt to trick the brain into perceiving an interaural time difference (ITD) and hence a direction.

An IntanRHD was connected to MATLAB, sounds were played and EEG signals (Cz and Pz) were recorded in sync. A small dataset was put together and was used to train simple models (multinomial regression, SVMs). The project was presented in the form of a game- MindSwiper. Over the course of 3 weeks, a model achieving up to 60% accuracies was trained.

Find the project repository here