| Summary: In the scope of the thesis, the words formed by the voluntary eye-blinks detected from the data acquired by using the EEG device have been transformed into speech. At the present time, there are many people in the world who have lost their ability to speak and move. These people are able to see and hear what is happening all around them, but they cannot communicate with the outside world. In addition, they cannot express their daily needs or thoughts to their environments. The aim of this thesis is to enable people to communicate their wishes through eyeblinks and thus to improve their quality of life. For this purpose, two different systems have been developed that convert Morse codes to speech. First, EEG signals were collected as a single channel from the electrode placed on the left eye with the wireless Neurosky MindWave device. In the collected EEG signals, there are signals originating from ocular movements except for brain activity. By using these signals, the words formed by Morse codes have been transformed into speech by the text to speech algorithm included in the Matlab. In the first system, after data collection; wavelet analysis, artificial neural network classification, Levenshtein distance and text-to-speech synthesis algorithms were used. In the second system, dynamic time warping and text-to-speech synthesis algorithms were applied to the data. The results obtained from both systems were compared, and very successful results were achieved especially in the second system. In the study also, single-channel EEG signal acquisition card with a low-cost, easy-to-use and highsensitivity was designed. Keywords: EEG, Brain-Computer Interface, Wavelet Transform, Classification, Levenshtein, Dynamic Time Warping, Eye-blink Detection. |
