Automatic Identification of Articulation Disorders for Arabic Children Speakers

Abualseoud Hanani, Mays Attari, Atta' Farakhna, Aseel Joma'A, Mohammed Hussein, Stephen Taylor
2016 Workshop on Child Computer Interaction  
Automatic identification of speech disorders in children's speech is very important for the diagnosis and monitoring of speech therapy. In this work, acoustic features (MFCC) have been used with the two most commonly used classification techniques in the speaker and language identification area, GMM-UBM and I-vector, for identifying three error types of speech production associated with phoneme [r] from Arabic children's speech. The sound [r] has been selected as it is the most common
more » ... ion problem that Arabic speaking children suffer from. The impact of [r] location in a word on the speech disorders has been investigated by considering words with [r] in the beginning, middle and end. The best performance of our 4-class systems, is 75% accuracy with our I-vector system and 61% for our GMM-UBM system. Performance of these two systems are improved to 92.5% and 83.4%, respectively, when the three disorder classes are combined into one class versus normal class (2-class systems).
doi:10.21437/wocci.2016-6 dblp:conf/wocci/HananiAFJHT16 fatcat:gqyrswrc7bexpcoxcbywteedt4