Analysis on Mel Frequency Cepstral Coefficients and Linear Predictive Cepstral Coefficients as Feature Extraction on Automatic Accents Identification

Automatic Accents Identification is very important for discussion especially within scope of speaker recognition. Some contribution of appropriate techniques uses in Music Recognition and Accent Identification may contributes in improving the recognition rate. Techniques in discussing on music genre identification or accents automatic identification and the combination of both processes still in ambiguous for this field. This paper investigates mainly the processes involved in speech processing or identification includes: acoustic/speech signal, pre-processing, feature extraction, pattern classification and accuracy results. Process of automatic accents identification through speech signals starts with general pre-processing techniques, feature extraction; which in this studies too, comparing within two techniques; Mel-Frequency Cepstral Coefficients (MFCC) and Linear Predictive Cepstral Coefficients (LPCC).While for vocal tract with musical characteristic in used for musical genre identification and the usage of pattern classification for three methods which includes; Hidden Markov Model (HMM), Support Vector Machine (SVM) and Probabilistic Principal Component Analysis. Thus, this paper investigates the feature extraction techniques used in identifying accents to be implemented in Quranic Accents identification and proposed MFCC as better techniques for feature extraction and getting higher accuracies for 93.33% while 86.67% if compared to LPCC.