Development of an Arabic speech database

Speech database is a backbone for developing any speech processing system or algorithm. For this reason we tried to set rules and specifications for Arabic database collection to be used in training and testing the Arabic speech processing system. These rules were adopted though studying Arabic language morphologically and linguistically in order to discriminate the shape and properties of each Arabic phonemes individually. A manual segmentation of raw speech record have been performed using the different shape of the Arabic phoneme as a guide line to extract and label the phoneme segment in order to built the Arabic speech database.