Abstract
Abstract Researches and techniques of pattern recognition aim at finding and developing techniques for recognizing certain patterns or structures in the digital signals. A signal can represent a sound segment or a verbal segment comprising a word or even a computer text. This research aims building a system that works on extracting the features distinguishing English letters pronunciation through entering sound segments (26 sound segments each of which representing one of the English letters) to the computer. (WAV) extension files have been used because this type of files cannot be affected by compression process. Each sound segment enters primary processing. After that, the distinctive features of each sound segment are extracted through finding the value of fractal dimension, by using the 2-Dimension Variation Algorithm, then analyzing and studying the fractal dimension values resulting from each sound segment to recognize the pronunciation of the entered letter. After, it has been shown that fractal dimension is a distinctive feature for pronouncing the entered English letters, and that using the 2-D Variation method for calculating fractal dimension yielded accurate results that comply with the research requirements.