Automatic Arabic Speech Segmentation Syste

growth of information and communication technologies has influenced the research trends n speech technologies. This research explains a basic speech segmentation application for Arabic language with the aim to further develop a language tutor. The focus is on rabic as there are standards available which help in obtaining better accuracy. The roblem has been formulated in the form of a number of possible cues that could help e phoneme boundaries. These cues include zero crossing rate (ZCR), power spectral SD), formant transitions, rhythm of consonants and vowels, intonation pattern also called al frequency and vowel duration. A number of experiments were carried out involving ombinations of cues. The best results for phoneme level segmentation were observed D and ZCR whereas pitch and intensity were helpful in determining pauses. Our system strated up to 89% accuracy on continuous speech files of eight different speakers which es approximately 14300 phonemes. : Speech Segmentation, Power Spectral Density, Zero Crossing Rate, Phoneme ion.