Voice Authentication Using Confidence-Weighted Word Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech verification systems are resource-intensive and require significant training to accurately identify speakers, making them less desirable for common authentication scenarios, and other authentication methods do not effectively utilize voice-based technologies, which are harder for impostors to imitate.
Innovation Solution
A computer-implemented method for voice-based authentication that compares voice input with a corpus of high-confidence and low-confidence words, using similarity scores and additional features like pitch, rhythm, and speaking speed to determine authenticity, allowing access to resources based on these criteria, even in multi-factor authentication processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speech verification is used to authenticate users, then speaker identity can be verified, but significant training and resources are required
Solution Approach 1:
The patent segments the authentication text into high-confidence words and low-confidence words, and segments the authentication process into multiple stages (initial authentication with high-confidence words, subsequent authentication with low-confidence words). This segmentation allows the system to achieve reliable speaker verification while reducing overall computational complexity by using different verification strategies for different word types.
Solution Approach 2:
The patent applies partial action by focusing verification efforts on specific subsets of words rather than treating all words equally. High-confidence words undergo rigorous verification, while low-confidence words are handled with simplified verification processes. This partial approach maintains authentication reliability for critical words while reducing overall system complexity.
2Reliability
If traditional authentication methods are used, then resource requirements are lower, but they do not leverage voice-based technologies which are harder for impostors to imitate
Solution Approach 1:
The patent applies local quality by treating different words in the authentication text differently based on their confidence levels. High-confidence words (which are harder to imitate) receive full verification scrutiny, while low-confidence words receive streamlined verification. This local differentiation maintains high security for critical authentication elements while improving overall authentication efficiency.
3Measurement precision
If all uttered words are verified with high confidence thresholds, then authentication accuracy increases, but false positives increase and user experience deteriorates
Solution Approach 1:
The patent dynamically changes the confidence threshold parameter based on the word type (high-confidence vs. low-confidence words). For high-confidence words, a strict threshold is applied to ensure accuracy. For low-confidence words, a more lenient threshold is used to maintain user experience. This parameter adaptation resolves the contradiction between precision and ease of operation.
Data Source
AI summary
A voice-based authentication system receives uttered words from a user (e.g., a human speaker); compares the uttered words with an authentication text that includes high-confidence corpus words and one or more low-confidence corpus words from previous training or authentication; identifies high-confidence uttered words and at least one low-confidence uttered word based on the comparison with the authentication text; compares the high-confidence uttered words with a threshold; determines that the at least one low-confidence uttered word corresponds to any of the low-confidence corpus words of the authentication text; and grants access to a resource (e.g., a user account, a document, a building, or a vehicle) based on the comparison of the high-confidence uttered words with a threshold and on the determination that the at least one low-confidence uttered word corresponds to any of the one or more low-confidence corpus words.