Abstract: A method and apparatus for processing data used in a character recognition system in which a threshold voltage value is developed to be applied to an analog waveform representing an unknown character to determine the start of the character waveform. Samples of the first peak of the waveform are generated and a value representing the area under the first peak is developed. Data processing means divides the area by the number of samples taken to arrive at a threshold voltage value which, when applied to the samples taken, determines the start of the character waveform. The voltage amplitudes of adjacent samples of the character waveform are then averaged and the number of samples reduced by one half. A predetermined number of samples having the maximum voltage amplitudes located within peak areas of the waveform are normalized for use by a utilization device in recognizing the unknown character.