Find in Library

This study focuses on the automatic decoding of inner speech using noninvasive methods, such as Electroencephalography (<span style="font-variant: small-caps;">EEG</span>). While inner speech has been a research topic in philosophy and psychology for half a century, recent attempts have been made to decode nonvoiced spoken words by using various brain–computer interfaces. The main shortcomings of existing work are reproducibility and the availability of data and code. In this work, we investigate various methods (using Convolutional Neural Network (<span style="font-variant: small-caps;">CNN</span>), Gated Recurrent Unit (<span style="font-variant: small-caps;">GRU</span>), Long Short-Term Memory Networks (<span style="font-variant: small-caps;">LSTM</span>)) for the detection task of five vowels and six words on a publicly available <span style="font-variant: small-caps;">EEG</span> dataset. The main contributions of this work are (1) subject dependent vs. subject-independent approaches, (2) the effect of different preprocessing steps (Independent Component Analysis (<span style="font-variant: small-caps;">ICA</span>), down-sampling and filtering), and (3) word classification (where we achieve state-of-the-art performance on a publicly available dataset). Overall we achieve a performance accuracy of 35.20% and 29.21% when classifying five vowels and six words, respectively, in a publicly available dataset, using our tuned iSpeech-CNN architecture. All of our code and processed data are publicly available to ensure reproducibility. As such, this work contributes to a deeper understanding and reproducibility of experiments in the area of inner speech detection.