Voice Authentication Threshold Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Voice authentication systems face challenges in setting appropriate thresholds to balance security and service levels, as different authentication engines produce varying scores for the same content types, leading to issues with false rejection and acceptance rates.
Innovation Solution
A method and system that systematically compare impostor voice samples against legitimate speaker voiceprints to derive authentication scores, analyze these scores to determine a measure of confidence, and set business rules based on the analysis, using an 'Impostor Map' to plot and analyze scores for optimal threshold settings across multiple authentication engines and content types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the threshold score is set too high to meet security requirements, then false acceptance rate decreases, but false rejection rate increases
Solution Approach 1:
The patent applies dynamics by making the threshold setting adaptive rather than fixed. The system dynamically adjusts threshold values based on analysis of authentication engine performance across different content types and impostor samples. This allows the threshold to optimize between security and service levels for each specific authentication scenario rather than using a static high threshold that works for all cases.
Solution Approach 2:
The patent changes the parameter of threshold values based on systematic analysis of authentication scores. By analyzing performance across multiple content types (account numbers, dates of birth, phrases) and comparing impostor versus legitimate speaker scores, the system determines optimal threshold settings for each content type, transforming a single fixed threshold into multiple adaptive thresholds.
2Ease of operation
If the threshold score is set too low to improve service levels, then false rejection rate decreases, but false acceptance rate increases
Solution Approach 1:
The system dynamically selects appropriate threshold values based on the specific content type being authenticated. For content types where authentication engines perform well (low false acceptance rates), the system can use lower thresholds to improve service levels. For content types with higher impostor success rates, the system uses higher thresholds to maintain security, thus optimizing the trade-off dynamically.
Solution Approach 2:
The patent changes threshold parameters based on systematic performance analysis. By comparing authentication engine scores across different content types and measuring impostor versus legitimate speaker differentiation, the system determines optimal threshold values for each content type, allowing low thresholds for secure content types while using higher thresholds for more vulnerable content types.
3Measurement precision
If different authentication engines use different attributes for voiceprint comparison, then measurement precision varies, but device complexity increases
Solution Approach 1:
The patent segments the authentication evaluation by content type (account numbers, dates of birth, phrases) and by authentication engine. This segmentation allows systematic comparison of performance across different engines and content types, identifying which engines perform best for each specific content type rather than trying to optimize all engines uniformly across all content types.
Solution Approach 2:
The patent applies local quality by optimizing authentication engine selection and threshold settings for each specific content type rather than using a uniform approach. Each content type receives customized authentication parameters based on its specific vulnerability profile, allowing maximum precision for each local case while managing overall system complexity through targeted optimization.
Data Source
AI summary
A method for configuring a voice authentication system employing at least one authentication engine comprises utilizing the at least one authentication engine to systematically compare a plurality of impostor voice sample against a voice sample of a legitimate person to derive respective authentication scores. The resultant authentication scores are analyzed to determine a measure of confidence for the voice authentication system.