Article
Creation of a Quranic Reciter Identification System using the GMM Classifier and MFCC
This research presents the design of Quranic Reciter Identification System based on Mel-Frequency Cepstral Coefficients (MFCC) Feature extract, and Gaussian Mixture Model (GMM) classifier. The algorithm aims at identifying each Quran reciter with a unique vocal characteristic. Spectral features of recitation audio are recorded by MFCC, which is effective in recording the tonalities of every voice of the reciter. The reciters are consequently grouped based on approximate probability distribution in voice pattern that is obtained by means of GMM, which models the retrieved attributes. Based on the outcomes of the experiment, in even alternative acoustic conditions, the system is capable of discriminating between many reciters, high levels of accuracy. The work contributes to the fields of speaker identification and Islamic audio processing and can be applied in archival recordings, verification of recordings and individual Quranic study.
Full Text Attachment
