Phoneme Sequence Matching for Voice Biometric Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice biometric systems for real-time fraud detection face challenges in accuracy and effectiveness due to limitations in text-independent voice biometrics, particularly with large voice print watchlists and high system processing loads.
Innovation Solution
The system employs a novel phoneme pattern comparison method, extracting and comparing phoneme sequences from voice prints to identify matches based on a similarity score, using techniques like Levenshtein distance calculation and vectorization to enhance accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If text-independent voice biometric systems are used for real-time fraud detection, then the system can process large numbers of voice prints, but the accuracy and effectiveness of fraud detection deteriorates due to high system processing loads and large watchlist sizes
Solution Approach 1:
The patent segments the voice print analysis into phoneme-level units, breaking down continuous voice signals into discrete phoneme sequences. This segmentation allows the system to compare specific phoneme patterns rather than processing entire voice prints, reducing computational complexity while maintaining detection accuracy. The phoneme sequences are extracted and compared independently, enabling efficient processing of large watchlists without sacrificing fraud detection precision.
2Reliability
If conventional voice print comparison methods are used, then the system can perform fraud detection, but false positives increase due to insufficient discrimination between similar voice patterns
Solution Approach 1:
The patent applies local quality by focusing comparison on specific phoneme sequences rather than treating all voice print portions equally. By identifying and comparing distinctive phoneme patterns that are characteristic of fraudster behavior, the system enhances matching precision. This localized approach to quality assessment allows the system to distinguish more accurately between genuine and fraudulent voice patterns, reducing false positives while maintaining reliable fraud detection.
Solution Approach 2:
The patent changes the parameter of comparison from whole-voice-print similarity to phoneme sequence matching. By transforming the comparison metric to operate at the phoneme level rather than the aggregate voice print level, the system achieves better discrimination between similar voice patterns. This parameter change enables more precise matching by focusing on specific acoustic features that are more discriminative for fraud detection.
Data Source
AI summary
A system and method for real-time fraud detection with a social engineering phoneme (SEP) watchlist of phoneme sequences may perform real-time fraud prevention operations including receiving incoming call interactions and grouping the call interactions into one or more clusters, each cluster associated with a speaker's voice based on voiceprints. For a pair of voiceprints in a cluster, a phoneme sequence is extracted for each voice print. From the extracted phoneme sequences, a similarity score is then calculated to determine if a match exists between the extracted phoneme sequences based on a threshold. If determined a match exists, the phoneme sequence may be added to a SEP watchlist.