Graph-Based Voice Authentication for Telephonic Interactions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems lack an organized method for selecting and using interactions for accurate agent and customer enrollment in voice print databases, leading to inefficient authentication processes.
Innovation Solution
A system and method that creates a graph to map interactions between agents and customers, identifies customers with the fewest interactions, selects interactions for agent enrollment, and uses other interactions for customer authentication, separating and comparing speaker components to generate and aggregate voice print scores for accurate enrollment and authentication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to select interactions for enrollment, then the process is simple, but the accuracy of voice authentication is insufficient
Solution Approach 1:
The patent segments the large set of customer-agent interactions into distinct categories using a graph-based approach. It identifies and separates interactions suitable for agent enrollment from those suitable for customer enrollment and authentication. This segmentation enables the system to select the most appropriate interactions for each purpose, improving voice authentication accuracy while maintaining a structured, manageable process.
Solution Approach 2:
The patent performs preliminary analysis and classification of interactions before the actual enrollment process. By creating a graph representation of interactions and identifying key characteristics in advance, the system prepares the data structure needed for accurate enrollment selection. This preliminary action ensures that when enrollment occurs, the most suitable interactions are already identified, improving both accuracy and efficiency.
2Reliability
If all interactions are used for enrollment, then the enrollment database is large, but the quality and accuracy of authentication decreases
Solution Approach 1:
The patent extracts and isolates specific interactions from the bulk of customer-agent communications that are most suitable for enrollment purposes. Using graph analysis, it identifies interactions with specific characteristics (such as clear audio quality, complete dialogue structure, and representative speech patterns) and separates them for dedicated enrollment use. This extraction ensures high-quality enrollment data without requiring all available interactions.
Solution Approach 2:
The patent applies different selection criteria to different portions of the interaction dataset. Rather than uniformly using all interactions, it identifies local regions or subsets of interactions that possess specific quality attributes ideal for enrollment. This local quality approach ensures that each enrollment record meets high standards while maintaining an appropriate quantity of diverse interactions in the overall database.
3Measurement precision
If a graph-based approach is implemented to map interactions, then enrollment accuracy improves, but system complexity increases
Solution Approach 1:
The patent introduces a graph-based representation as an intermediary structure between the raw interaction data and the enrollment decision-making process. This graph model serves as a mediator that organizes interactions according to their relationships and characteristics, making it easier to identify suitable enrollment candidates. The graph structure provides a systematic way to analyze and select interactions without requiring complex algorithms at every stage.
Solution Approach 2:
The patent transforms the traditional linear or tabular representation of interactions into a graph-based dimensional structure. This dimensional change allows the system to capture relationships and patterns that are not apparent in conventional data formats. By organizing interactions in a graph dimension with nodes and edges representing different aspects of the interactions, the system achieves more precise enrollment selection through enhanced data organization rather than increased computational complexity.
Data Source
AI summary
Methods for voice authentication include receiving a plurality of mono telephonic interactions between customers and agents; creating a mapping of the plurality of mono telephonic interactions that illustrates which agent interacted with which customer in each of the interactions; determining how many agents each customer interacted with; identifying one or more customers an agent has interacted with that have the fewest interactions with other agents; and selecting a predetermined number of interactions of the agent with each of the identified customers. In some embodiments, the methods further include creating a voice print from first and second speaker components of each interaction; comparing the voice prints of a first selected interaction to the voice prints from a second selected interaction; calculating a similarity score between the voice prints; aggregating scores; and identifying the voice prints that are associated with the agent.