Introduction of Beat Oscillation to Improve the Performance of Music BCI Decoder

Youmin Shin, Jii Kwon, June Sic Kim, Chun Kee Chung

Abstract: A Brain-Computer Interface (BCI) is emerging as next-generation technology to overcome disabilities and human physical limitations. Recently, the BCI for language and music as well as physical activities are being developed, which improves accessibility in communication and creative activities. BCI technology requires high accuracy, but despite many engineering approaches, it still does not have sufficient performance. In this study, we tried an approach based on the brain mechanism. There is a limitation due to high inter-subject variability. Also, Brain waves represent responses to various activities such as cognitive processes, visual information, and body movement. Therefore, it is important to design a personalized classifier and extract only responses to specific activities. Beat entrainment is known as the key mechanism of the human brain. Beat entrainment is in charge of segmentation on external information and is greatly involved in high-cognitive functions such as language, movement, and music. Introducing beat oscillation in the decoder, we confirmed that BCI performance increased. We recorded the brain response through the ECoG electrode while listening to the familiar children's song. We designed a classifier based on a 7-tone diatonic scale (Do-Re-Mi-Fa-Sol-La-Ti) relative to the tonic (Do). The accuracy was significantly higher in the beat-applied classifier than otherwise.