Voice Authentication Watermark Embedding for Security
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Voice authentication systems face security issues due to unauthorized access and forgery, particularly in medical record management, where recorded voices can be used to impersonate users, compromising the reliability and integrity of speaker authentication.
Innovation Solution
A voice authentication system incorporating a deep neural network (DNN) model to generate a voice image, extract feature vectors, embed watermarks with individual information, and use encryption to ensure secure access, preventing unauthorized modifications and forgery by embedding watermarks into voice data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If voice authentication is performed using recorded voice data, then authentication speed is improved, but security is worsened due to potential forgery and unauthorized access
Solution Approach 1:
The system performs preliminary actions by extracting feature vectors from voice data during the registration phase and storing them securely. During authentication, these pre-extracted feature vectors are compared against the new voice input, enabling fast authentication while maintaining security through the use of encrypted and watermarked feature data.
Solution Approach 2:
The patent introduces watermarks as an intermediary mechanism embedded in the voice authentication data. These watermarks act as a mediator between the voice data and the authentication system, allowing the system to verify authenticity and detect forgery without slowing down the authentication process.
2Measurement precision
If deep neural network is used for voice recognition, then authentication accuracy is improved, but device complexity is worsened
Solution Approach 1:
The system extracts only the essential feature vectors from the complex deep neural network processing of voice data. By taking out and storing these condensed feature representations during registration, the system maintains high authentication accuracy while reducing the computational complexity required during subsequent authentication operations.
3Reliability
If watermark embedding is applied to voice data, then data integrity is improved, but processing time is worsened
Solution Approach 1:
The watermark embedding is performed as a preliminary action during the registration phase when voice data is first collected and processed. By embedding watermarks in advance into the feature vectors and voice images, the system ensures data integrity without adding processing time during subsequent authentication operations, as the watermarks are already in place for verification.
Data Source
AI summary
The present disclosure provides a voice authentication system. The voice authentication system according to an embodiment of the present disclosure includes a voice collection unit configured to collect voice information obtained by digitizing a speaker's voice, a learning model server configured to generate a voice image based on the collected voice information of the speaker, causes a deep neural network (DNN) model to learn the voice image, and extract a feature vector for the voice image, a watermark server configured to generate a watermark based on the feature vector and embed the watermark and individual information into the voice image or voice conversion data, and an authentication server configured to generate a private key based on the feature vector and determine whether to extract the watermark and the individual information based on an authentication result.