Voice Authentication Recursive Fragment Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current authentication methods in IP networks require additional non-biometric features due to the computational intensity of comparing biometric data, such as voiceprints, with large reference datasets, making it impractical to authenticate users solely based on their voice in real-time.
Innovation Solution
A method that processes voice inputs by transforming them into audio fragments, analyzing, and comparing these fragments with a reference database using a speech processing platform, calculating confidence values, and recursively reducing the comparison set to identify a matching voiceprint, allowing for efficient authentication without additional features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If biometric data comparison is performed with large reference datasets, then authentication accuracy is improved, but computational complexity and processing time increase
Solution Approach 1:
The patent segments the large reference dataset into multiple smaller subsets organized in a hierarchical structure. The comparison process is divided into stages where audio fragments are first compared against coarse-grained subsets, then progressively refined to finer subsets. This segmentation reduces the computational complexity of comparing against the entire dataset while maintaining authentication accuracy through multi-stage verification.
Solution Approach 2:
The patent performs preliminary organization of reference biometric data into hierarchical subsets before the actual authentication process. Reference audio fragments are pre-grouped by similarity metrics and stored in an organized structure that enables efficient retrieval. This preliminary action allows the system to quickly eliminate large portions of the reference dataset that don't match the query, reducing real-time computational requirements.
2Measurement precision
If biometric data comparison is performed with large reference datasets, then authentication accuracy is improved, but processing time increases
Solution Approach 1:
The patent divides the reference dataset into hierarchical subsets and processes comparisons in stages rather than performing a single exhaustive comparison. This segmentation enables the system to quickly eliminate non-matching references in early stages and only perform detailed comparisons on a small subset of potential matches, dramatically reducing total processing time while maintaining accuracy.
Solution Approach 2:
The patent implements a multi-stage comparison process where not all reference fragments are compared with equal depth. Initial stages use coarse matching with lower computational effort to filter out obvious non-matches, then subsequent stages apply more rigorous comparison only to promising candidates. This partial action approach achieves sufficient accuracy without the excessive time cost of comparing every reference fragment in detail.
3Ease of operation
If voice-only authentication is implemented, then ease of operation is improved, but reliability decreases due to computational limitations
Solution Approach 1:
The patent segments the authentication process into multiple verification stages, each operating on a subset of reference data. This segmentation allows the system to maintain high reliability by performing thorough verification on relevant subsets while keeping the overall process simple and voice-only for the user. The hierarchical structure ensures that no single comparison point fails the entire authentication.
Data Source
Figure 1
AI summary
The present invention describes a method for identifying and authenticating a user (10) in order to grant the user (10) access to a secure area in an IP network, in particular to confidential data and/or services, wherein, by means of an audio-enabled device (20), a speech input spoken by the user (10) into the audio-enabled device (20) is received and transformed into a sequence of audio fragments, and in a speech processing platform (3) each audio fragment is compared with a set of reference speechprints stored in a reference database that potentially match the speechprint in order to find a matching reference speechprint, wherein each stored reference speechprint that does not sufficiently match is discarded from the set of potentially matching reference speechprints, and the comparison (26,36) For each subsequent audio fragment in the sequence, a recursive analysis is performed using only the potentially matching reference language prints remaining in the set from the reference database until only one reference language print remains as a possible matching reference language print. Furthermore, a corresponding system and speech processing platform are provided.