Abstract
An experimental evaluation of an automatic algorithm for segmentation of speech into Arabic syllabic units is reported here. The parameter used for segmentation is the energy and statistical parameter of acoustic signal and depending on the characteristics of the phonology in Arabic language {and the fairly close correspondence between the acoustics and phonetics of these syllables}. Speech data of four speakers (2 male, 2 female), consisting of mono-syllabic and poly-syllabic words were used to test the segmentation algorithm. It is shown that the final output result of this test is valid for a bout 85%.