A Novel Auditory BCI for Spelling Persian Words

Abstract

In this paper a novel auditory BCI paradigm based on P300 responses which are generated by auditory stimuli is proposed. In the proposed protocol, stimuli are different from each other and there is no need to repeat each single stimulus. Instead of that, a 30-word list containing different words as stimuli is played for the subject. Each letter is repeated in four different words of the list. Ensemble support vector machine and regularized linear discriminant analysis classifiers are used to evaluate the proposed protocol. The result of the offline classification of the P300 responses of one subject shows we can achieve accuracy and ITR of 88% and 3.51 bit/min for three iterations of the list (twelve repetitions of a letter).