Comparative Study of EEG-based Speech Detection with Novel Feature Set

Building upon previous research in non-invasive speech detection using brain activity, this study aimed to verify the performance of a previously proposed algorithm with a new dataset. This dataset offered greater variability in participants and mental tasks compared to the original dataset. Verification of the algorithm algorithm achieved an accuracy of 84% in distinguishing pronounced speech, inner speech, and silence intervals. This verification using a different dataset strengthens the algorithm’s potential for various applications. Developing algorithms for non-invasive speech detection using brain activity holds significant potential for communication assistance and understanding human cognition. Further extending the algorithm could emphasize the potential of developing EEG-based speech recognition.