[3J1-GS-1] Fundamental AI, theory

[3J1-GS-1-03] Extracting linguistic representation from speech-imagery EEG.

〇Tsuneo Nitta¹, Junsei Horikawa¹, Takuro Fukuda², Kouichi Katsurada², Daisuke Suzuki³, Yurie Iribe³, Ryo Taguchi⁴, Shuji Shinohara⁵, Goh Kawai⁶ (1. Toyohashi University of Technology, 2. Tokyo University of Science, 3. Aichi Prefectural University, 4. Nagoya Institute of Technology, 5. Tokyo Denki University, 6. Tokyo University of Foreign Studies)

Keywords:Brain computer interface, EEG signal, linguistic representation

Speech imagined signal recognition from Electroencephalogram (EEG) is one of the challenging technologies for non-invasive brain-computer-interface (BCI). In this report we regard the representation of language as the difference of line spectra of syllables observed around Broker area and develop the method for extracting line spectra in EEG signal. The BCI research framework based on the extraction of language-representation, as well as vowels and consonants recognition using twenty one electrodes around Broker area, are presented.