International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022
p-ISSN: 2395-0072
www.irjet.net
Voice-based Login System Vedant Baviskar1, Shubham Bhagwat2, Deepjyoti Barman3, Gokul Viswanath4, Mrs. R.T. Waghmode 5 1,2,3,4,5 B.Tech
student, Computer Science, Sinhgad Institute of Technology and Science, Pune, India ---------------------------------------------------------------------***--------------------------------------------------------------------1.1 RELEVANCE Abstract - This project on Voice Authentication analyses audio to determine the identity of a user and authenticate them into a system while using Gaussian Mixture Models. This has a wide set of uses, from a normal web application to a high-security facility.
There are many areas that can use Biometric identification (in this case, Voice Authentication). The main areas this can be used are the ones where security is prioritized, ranging from a company login for admin tools to a high-security military checkpoint to watch border crossings at different frontiers.
Understanding how complex and high dimensional audio can be recognized is one a challenge. Audio data contains information that can be used effectively for information or can be filtered out (in the case of our project) using Audio Content Analysis. This information includes background noise, crowd chatter and other noise like animals or cars. These reasons make audio recognition and classification a very challenging research topic. High dimensional audio can be exceedingly large in size and can pose problems to the speed of authentication and the performance impact on the devices used. So to solve this problem, we use audio formats such as MPEG to maintain the high level of detail in the audio clip, while also keeping the file size small.
Higher ranked military forces can use this to regulate entry to particular top-level facilities. Although the military is one main sector, even normal social media logins can be done using voice authentication, if they wish to. When a person is driving a vehicle, they can use systems that use Voice Authentication and recognition to use services while also keeping their eyes on the road, thus contributing to road safety.
Key Words: High dimensional audio, Audio Content Analysis, Moving Pictures Expert Group, Gaussian Mixture Model
People with dyslexia would experience difficulties reading items on a screen and thus can use Voice Authentication to log in and use a service without much difficulty.
1. INTRODUCTION
1.2 MOTIVATION
Voice Biometrics is a technology that relies on the recognition of voice patterns to verify the identity of individuals. This is possible as each person’s vocal tract is unique. Physical features, both phonetic and morphological are unique to each individual making them virtually fraud proof. It enables fast, frictionless and a very secure way to authenticate users.
As stated before, analysing and understanding multidimensional audio data is a big challenge. Using data from that analysis and implementing it in a real-time system is even more so. Such analyses and implementations can provide useful information to us about audio, and how they can be used in a wide range to applications that we use every day as part of our lifestyle. The voice is something you are, and uniquely defines you and your characteristics. This can help people and organizations introduce more such platforms that use Voice Authentication, and thus make the users' life easier while maintaining (or even strengthening) the level of security.
This process of authentication works in two steps, the first being the classification of the audio into speech and nonspeech audio, the latter being music, noise and environment sounds. The second part starts by creating a list of characteristics that the trained model perceives from the voice, which is stored in an array. Then the user's characteristics that were saved during registering will be downloaded and used to compare both arrays. If the distance between both the arrays is below a specific threshold value, then the user is authenticated.
© 2022, IRJET
|
Impact Factor value: 7.529
2. REQUIREMENTS Following are the requirements for the Voice Authentication system.
|
ISO 9001:2008 Certified Journal
|
Page 538