A REVIEW ON AUDIO AND VIDEO BASED PERSON AUTHENTICATION USING MACHINE LEARNING

Computer Engineering

Identity verification is a method that usually uniquely identifies a person based on a password and a personal identification number (PIN). The main problem with this authentication technology is that users are not willing to remember long-term and challenging combinations of numbers, letters and symbols that may be lost, forged, stolen or forgotten. In this article, we examined recent developments in the use of behavioral biometrics for user authentication. The application of behavioral biometric authentication basically consists of three main modules, namely data capture, function extraction and classification. The application focuses on extracting user-related behavioral characteristics and using these properties for identity verification measures. The goal is to determine the classification technology that is primarily used for data analysis in the identity verification process. Through comparison, we expect to find gaps in improving the performance of behavioral biometric authentication. In addition, we highlight a set of classification technologies that are best suited for behavioral biometric approval. This article discusses the concept of speech recognition in deep learning methods. This article will discuss speech recognition, deep learning and the introduction of deep learning methods. This article also describes models for deep learning for speech recognition. This article defines the work related to speech recognition using deep learning methods as well as the software related to Sphinx, which allows speech recognition in Java languages. The main purpose of this review is to define the use of the sphinx and solar eclipse to recognize speech. We would suggest that we have proposed a new human recognition technology using face and voice fusion that can greatly improve the recognition speed compared to a single biometric recognition used in the development of security systems. Our system uses the Viola Jones algorithm for face recognition. The proposed system uses the Local Binary Pattern (LBP) as a technique for extracting functions to calculate local functions. In our project, MFCC function (Extracting) with Mel Frequency Divergence Coefficient (technology for extraction) is used for speech recognition. The extracted function provided by the SVM classification as input can be used to identify individuals and then display the results. The new system can be used in a variety of areas, such as identity verification and other potential commercial applications.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.