Origin of the data These Audio files are related to the publication "A dataset of EEG and EOG from an auditory EOG-based communication system for patients in locked-in state" by Andres Jaramillo-Gonzalez, Shize Wu, Alessandro Tonin, Aygul Rana, Majid Khalili-Ardali, Niels Birbaumer, and Ujwal Chaudhary. To use the audio data, please contact Dr. Ujwal Chaudhary, as mentioned in the manuscript. Details of the data The description of the nature and details of the *.txt files are in the section "Data Records" of the manuscript. As mentioned in the manuscript, before the beginning of the study, at least 100 questions with known "yes" or "no" answers were formulated and recorded with a family member or caretaker's voice close to the patient. Each question with a "yes" answer is paired with a similar question with a "no" answer (e.g., "Paris is the capital of France" and "Berlin is the capital of France"). Each question is saved as an audio file with an explicit identifier, a question with a "yes" answer is saved with a 001_NUMBER identifier, and a question with a "no" answer is saved with a 002_NUMBER identifier. The NUMBER label is composed of 5 digits; the first two determine the number of the patient, and the last three digits indicate the number of the question. The value of the label NUMBER is the same for a semantically paired sentence. In the example, "001_11007.wav" indicates a "yes" type of answer, question number 07 from P11. Sentences are then placed in a specific folder of the used laptop's storage, accessed and played by the communication system along with the sessions during sentence presentation.