Scam Call Detection via Voice Biometric Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging for individuals to determine whether incoming voice calls are scams, as many scam calls appear legitimate and can be difficult to identify during the call, especially when the caller's number and voice are not easily verifiable in real-time.
Innovation Solution
A computing system that allows users to evaluate the risk of a scam call during a voice call by recording the caller's number and voice sample, comparing them to a database of known scam callers, and providing an alert to the user without interrupting the call, using methods such as frequency estimation or hidden Markov models to determine the scam risk.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real-time analysis of caller voice and number is performed during a voice call, then scam risk detection capability is improved, but call processing time and system complexity increase
Solution Approach 1:
The system performs preliminary actions by maintaining pre-trained voice biometric models and scammer voice profiles in the database before calls occur. During a call, the system simply compares the caller's voice against these pre-established references, avoiding the need for complex real-time analysis algorithms. This preliminary preparation of reference data enables fast, reliable scam detection without adding significant computational complexity during the actual call processing.
2Measurement precision
If voice sample analysis is conducted during an active call, then accuracy of scam identification is improved, but call interruption and user experience deteriorate
Solution Approach 1:
The system replaces mechanical interruption methods (such as hanging up the call or displaying blocking UI elements) with electronic signal substitution. Instead of interrupting the call flow, the system subtly embeds analysis requests within the ongoing audio stream and delivers results through non-intrusive electronic notifications. The voice biometric analysis operates on captured audio samples without requiring call termination or user interaction, maintaining call continuity while achieving accurate scam identification.
3Reliability
If caller voice samples are recorded and analyzed, then ability to detect scam calls is improved, but privacy concerns and data security requirements increase
Solution Approach 1:
The system extracts only the essential voice biometric features (such as pitch, timbre, and speech patterns) from the caller's voice sample, separating these analytical characteristics from the actual voice recording. Instead of storing or transmitting the complete voice sample, the system processes and retains only the extracted biometric markers needed for comparison against scammer profiles. This extraction approach maintains scam detection reliability while minimizing privacy intrusion and data security requirements.
Data Source
AI summary
Examples are disclosed that relate to detecting and providing alerts regarding a possible scam voice call. One example provides a method comprising receiving a voice call, and obtaining, during the voice call, a caller number and a caller voice sample. The method further comprises obtaining, via the caller number and the caller voice sample, a risk determination regarding the risk that the voice call is a scam call, and based upon the risk determination, outputting, during the voice call, an alert regarding the risk that the voice call is a scam call.