Intelligent System for Speaker Identification using Lip features with PCA and ICA
Biometric authentication techniques are more consistent and efficient than conventional authentication techniques and can be used in monitoring, transaction authentication, information retrieval, access control, forensics, etc. In this paper, we have presented a detailed comparative analysis between Principle Component Analysis (PCA) and Independent Component Analysis (ICA) which are used for feature extraction on the basis of different Artificial Neural Network (ANN) such as Back Propagation (BP), Radial Basis Function (RBF) and Learning Vector Quantization (LVQ). In this paper, we have chosen "TULIPS1 database, (Movellan, 1995)" which is a small audiovisual database of 12 subjects saying the first 4 digits in English for the incorporation of above methods. The six geometric lip features i.e. height of the outer corners of the mouth, width of the outer corners of the mouth, height of the inner corners of the mouth, width of the inner corners of the mouth, height of the upper lip, and height of the lower lip which extracts the identity relevant information are considered for the research work. After the comprehensive analysis and evaluation a maximum of 91.07% accuracy in speaker recognition is achieved using PCA and RBF and 87.36% accuracy is achieved using ICA and RBF. Speaker identification has a wide scope of applications such as access control, monitoring, transaction authentication, information retrieval, forensics, etc.