Abstract

An automated computerized system to convert Arabic graphemes to phonemes (G2P) by using Arabic language phonology rules supported by a dictionary of exceptional words. The system was tested on a publicly dataset that contains 620 fully diacritics Arabic sentences formed from 3440 words and consists of 27030 graphemes which were manually segmented. the results were very promising on different levels of validation: sentences, word and phonemes. The system showed a high rate of accuracy and the precision of the system is 99.19%; with sensitivity rate of 99.42%. All Arabic rules were applied and tested. The developed system can be applied to any diacritic Arabic text.