an electromyogram signal obtaining unit including electromyogram sensors disposed at both sides for generating an electromyogram signal according to a predetermined muscle activated when a user clenches teeth, and a ground electrode connected to a body of the user for providing a reference voltage; and

an electromyogram signal processing unit for outputting a character by identifying a teeth clenching motion pattern of each block of teeth according to a side of teeth clenched, a duration of clenching teeth and consecutive clenching motions through analyzing the obtained electromyogram and selecting characters according to the identified teeth clenching motion pattern,

wherein the electromyogram signal obtaining unit comprises:

a basic pattern identifying unit for dividing the electromyogram signal by a predetermined time, extracting characteristics from the divided electromyogram time blocks, comparing the extracted feature value with a predetermined threshold value, classifying the divided electromyogram time block into an On-state if the extracted feature value are larger than the threshold, classifying the divided electromyogram time block into an Off-state if the extracted feature value is smaller than the threshold, and identifying a basic teeth clenching motion pattern having one of values ‘rest’, ‘left’, ‘right’ and ‘both’; a command generating unit for selecting a key location allocated on a keyboard through identifying a motion of clenching left teeth shortly, a motion of clenching right teeth shortly, a motion of clenching left teeth long and a motion of clenching right teeth long from a basic teeth clenching pattern of each block, and selecting one of characters in the selected key by clenching both side teeth shortly, clenching both side teeth twice and clenching both side teeth long; and

a character displaying unit for displaying the keyboard, and outputting characters outputted from the command generating unit in video and audio, and

wherein the dividing the electromyogram signal by a predetermined time is configure to obtain a difference of absolute means value (DAMV) as feature value from the divided electromyogram signals obtained from two channels by DAMV=1N−1.times.i=2N.times.x.function.(i)−x.function.(i−1), where N denotes the number of samples in each of the divided electromyogram signals and x(i) represents the digital electromyogram signal value at an index i.