Real Time Speaker Identification Using PLP & MFCC Feature Extraction Methods and SVM & GMM-UBM Classification Model
The aim of this study was to compare the performance of two feature extraction methods, namely Perceptual Linear Prediction (PLP) and Mel-Frequency Cepstral Coefficients (MFCC), in combination with two classification algorithms, Support Vector Machines (SVM) and Gaussian Mixture Model-Universal Background Model (GMM-UBM), for real-time speaker identification. The study uses a dataset consisting of 9 recordings of students as training & test sets to train and test the system. Results show that the combination of MFCC and SVM achieves the highest accuracy of 83.5% on the test set.