Survey of Features Extraction and Classification Techniques for Speaker Identification
Keywords:Speaker Recognition, Vector quantization, Feature Extraction, MFCC, Classifiers
Speech processing is more common day by day to provide enormous safety. The speech for the purpose of authentication is commonly used. Recognition of the speaker is the method that can check and recognize the speaker. The scheme of speech recognition is distinct from the scheme of speaker recognition. Recognition of speakers is commonly used in sectors, hospitals, laboratories, etc. Its benefits are safer, easier to implement, more user-friendly. Speaker identification method is one of the most commonly used techniques for the region where safety is very crucial. This article presents an overview of various methods that can be used to recognize speakers’ systems, the feature extraction techniques such as Linear Predictive Coding (LPC), Linear Predictive Cepstral Coefficients (LPCC), Unique Mapped Real Transform (UMRT), Real Cepstral Coefficients (RCC), “Mel-frequency Cepstrum” (MFCC), in addition to various classification techniques such as “Gaussian mixture model (GMM)”, “Dynamic Time Warping (DTW)”, Support Vector Machine (SVM), Neural Network (NN), “Vector Quantization” (VQ). The primary purpose of is to explain the common speaker recognition methods. The obtained results are that, MFCC is chosen for high efficiency and low complexity. and GMM is helpful in classifying less memory and less planning and efficient test results.
How to Cite
Copyright (c) 2020 Journal of University of Babylon
This work is licensed under a Creative Commons Attribution 4.0 International License.