To search, Click below search items.


All Published Papers Search Service


A Complete Mispronunciation Detection System for Arabic Phonemes using SVM


Muazzam Maqsood, Hafiz Adnan Habib, Tabassam Nawaz, Khurram Zeeshan Haider


Vol. 16  No. 3  pp. 30-34


Computer Assisted Language Learning Systems have gained a lot of attention in recent decades. Mispronunciation detection is probably the most important feature of these systems. It helps user to find out their pronunciation mistakes and provide useful feedback related to that mistake. Mispronunciation detection systems can be categorized in two classes Posterior Probability based and Classifier based systems. In this paper pronunciation assessment problem is formulated as a classification problem. This research paper explores the Acoustic Phonetic Features (APF) rather than traditional Confidence Measure based scores for mispronunciation detection. Support Vector Machines (SVM) is used as a classifier to detect pronunciation mistakes. As a test case five Arabic phoneme are tested for mispronunciation detection. APF based classifier produced excellent results and give average accuracy of 97.5%. The proposed system outperforms the existing systems that have been developed for Arabic phonemes.


Computer Assisted Language Learning Systems, Mispronunciation Detection, SVM, Acoustic Phonetic Features