Computer Aided Pronunciation Learning System Using Speech Recognition Techniques

This paper describes a speech-enabled Computer Aided Pronunciation Learning (CAPL) system HAFSS. This system was developed for teaching Arabic pronunciations to non-native speakers. A challenging application of HAFSS is teaching the correct recitation of the holy Qur'an. HAFSS uses a state of the art speech recognizer to detect errors in user recitation. To increase accuracy of the speech recognizer, only probable pronunciation variants, that cover all common types of recitation errors, are examined by the speech decoder. A module for the automatic generation of pronunciation hypotheses is built as a component of the system. A phoneme duration classification algorithm is implemented to detect recitation errors related to phoneme durations. The decision reached by the recognizer is accompanied by a confidence score to reduce effect of misleading system feedbacks to unpredictable speech inputs. Performance evaluation using a data set that includes 6.6% wrong speech segments showed that the system correctly identified the error in 62.4% of pronunciation errors, reported "Repeat Request" for 22.4% of the errors and made false acceptance of 14.9% of total errors.