Multicommand Auditory ERP-based BCI System

Abstract

This paper proposes an auditory modality-based braincomputer interface (BCI) using electroencephalogram (EEG) signals. The objective is an alternative BCI method for control and communication. The N200 and P300 responses from vowel sound stimuli were observed in a preliminary study to repeat and verify an auditory stimulus method. By using a single speaker and the proposed paradigm, a simple classification method can produce four commands. The results showed that the proposed auditory stimulation of a vowel sound duration of 250 ms with 20 trials of stream stimuli achieved a high average accuracy; the average accuracy of combining the ERP features produced from the Fz, Cz, Pz, P3, and P4 channels was 78.13%. We will employ the proposed auditory modality for a real-time BCI system to serve paralyzed patients with visual impairments.