Voice Authentication Using Caller ID to Reduce Computational Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The computational intensity of comparing speech samples with a large number of voice prints in voice-based speaker recognition and authentication systems leads to inefficient and costly identification processes for a large number of users.
Innovation Solution
The method involves using Network Provided CLI (Customer Line Identification Number) to reduce the number of voice prints that need to be compared by linking caller identification information with speaker feature information, allowing for faster and more efficient identification and authentication by checking if the recorded speech information matches the stored speaker feature information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If speech samples are compared with all available voice prints to ensure accurate speaker recognition, then identification accuracy is improved, but computational time and processing costs increase significantly
Solution Approach 1:
The system pre-processes speech samples during telephone connections to extract speaker feature information and creates voice prints in advance. This preliminary action stores essential biometric data that can be quickly retrieved and compared later, eliminating the need for real-time extensive processing during authentication events.
Solution Approach 2:
The invention extracts and stores only the essential speaker feature information from complete speech samples. By taking out and storing only the critical biometric characteristics (voice prints) rather than entire speech recordings, the system reduces data volume for comparison while maintaining identification accuracy.
2Reliability
If speech samples are compared with a large number of voice prints to authenticate users, then security is improved, but processing costs and system complexity increase
Solution Approach 1:
Voice prints are created and stored in advance during telephone connections before authentication is needed. This preliminary creation of biometric templates allows the system to maintain high security through comprehensive voice print databases without requiring complex real-time processing infrastructure during authentication events.
Solution Approach 2:
The system creates simplified representations (voice prints) of speech samples that capture essential biometric features. These copied feature sets retain the security value of original speech data while being computationally efficient for storage, retrieval, and comparison operations.
3Reliability
If comprehensive speaker recognition is performed for all users, then identification reliability is improved, but contact time with the service increases
Solution Approach 1:
Speaker feature information is extracted and voice prints are generated during routine telephone connections when users are already engaged with the service. This preliminary processing ensures that when authentication is required, the system can quickly retrieve and compare stored voice prints without extending user contact time.
Solution Approach 2:
The system automatically extracts speaker features and creates voice prints during telephone connections without requiring additional user actions or extending service interaction time. The authentication process serves itself by utilizing existing communication sessions for biometric data collection.
Data Source
Figure 1~2
AI summary
A method is described for improved identification and/or authentication of a user during a telephone connection or voice call with a voice telephony system, wherein at least one speaker feature information together with at least one caller ID information is available or accessible for a plurality of users, and wherein the method comprises the following steps: -- in a first step, a telephone connection or voice call is established to the voice telephony system from a specific telecommunications terminal of a specific user, wherein a specific caller ID information is assigned to the specific telecommunications terminal and wherein the specific caller ID information is transmitted to the voice telephony system when the telephone connection or voice call is established, -- in a second step following the first step, it is checked whetherWhether a specific speaker characteristic information exists or is accessible with respect to the specific caller ID information – in a third step following the second step, if the check to determine whether a specific speaker characteristic information exists or is accessible with respect to the specific caller ID information is positive, it is checked whether voice information from the user, captured during the duration of the telephone connection or voice call, shows or yields a sufficiently high degree of similarity with the specific speaker characteristic information, whereby automated identification and/or authentication of the specific user takes place if such a sufficiently high degree of similarity is found.