International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 05 | May 2024
p-ISSN: 2395-0072
www.irjet.net
Voice Biometrics – A New Outlook to a Traditional Problem Nakshatra Joshi1, Mohit Ghadi2, Sahil Kumar3, Swati Nandusekar4, Sunil Patil5 1Student, Dept. of AI & DS Engineering, KJ Somaiya Institute of Technology, Maharashtra, India 2Student, Dept. of EXTC Engineering, KJ Somaiya Institute of Technology, Maharashtra, India 3Student, Dept. of EXTC Engineering, KJ Somaiya Institute of Technology, Maharashtra, India
4Assistant Professor, Dept. of AI & DS Engineering, KJ Somaiya Institute of Technology, Maharashtra, India 5Assistant Professor, Dept. of EXTC Engineering, KJ Somaiya Institute of Technology, Maharashtra, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The traditional method of authentication using a
We have adopted Gaussian Mixture Models (GMM) and Mel Frequency Cepstral Coefficients (MFCC) features as the pillars of our method of building the voice-based authentication system. MFCC features enable us to transform the voice input into a compact representation which encapsulates the unique characteristics of an individual's speech. These features are able to capture the subtle features that define an individual’s identity. GMM models are a powerful tool for statistical modelling and classification, which are extremely well suited for modelling a person’s voice. This helps us create robust templates for genuine user verification while effectively detecting unauthorized attempts. Through the integration of MFCC features and GMM models, our approach aims to build a robust voicebased authentication system.
one-time password does not provide enough security for web pages or applications that are a gateway to sensitive information/data. Sensitive data breaches have become a common thing in today’s world and have caused great losses to individuals, organizations and nations as well. With advancement in technology, access to OTP has become a piece of cake and sensitive data has become vulnerable. Given the gravity of the situation, a more secure way of authentication has become the need of the hour, therefore voice-based authentication is a better successor to the traditional method and is more secure. In this paper we have collected the voice sample and processed it using the MFCC method to extract the voice print and then used Gaussian Mixture Model to find similarity between the stored voice and the live voice.
A. Components:
Key Words: Voice Authentication, MFCC, GMM, Sensitive data breaches, Voice print
In a voice-based authentication system, user verification and user registration are two of its important components. When combined, these two elements form a dependable and efficient authentication mechanism.
1.INTRODUCTION In today’s world, new cyber-attacks happen every single day, which has resulted in the need for robust and user-friendly authentication systems. Traditional methods of authentication include passwords, PINs and OTPs. These conventional methods are vulnerable to hacking attempts using methods such as social engineering, phishing, and brute force assaults. Voice-based authentication systems offer an excellent solution to address the drawbacks of these methods. It is a biometric method of authentication that makes use of the uniqueness of an individual’s voice, as no two people have the same voice, no matter how similar they sound. By utilising distinct vocal characteristics of individuals, including intonation, rhythm, and pronunciation, voice-based authentication systems provide a highly personalized and secure means of identity verification. These systems do not require any fancy hardware, unlike other authentication systems, they only require a regular microphone which is built into almost every device like mobile phones and laptops nowadays. This is why this system holds promise in diverse applications, ranging from mobile devices and smart home assistants to enterprise-level security protocols.
© 2024, IRJET
|
Impact Factor value: 8.226
Components of Voice Authentication
User Registration
User Verification
Fig -1: Components of Voice Authentication
|
ISO 9001:2008 Certified Journal
|
Page 571