Acoustic Call Verification Against DTMF Emulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing IVR systems are vulnerable to fraudsters using autodialers to circumvent confirmation schemes, allowing fraudulent calls to continue and generate unnecessary fees by mimicking live callers through DTMF code emulation, leading to unwanted robocalls, toll-free pumping, and other phone frauds.
Innovation Solution
Implementing a system that holds incoming calls for a short period to analyze audio patterns against predefined fraudulent patterns, routing suspicious calls to an IVR for code input, and terminating calls that fail to enter the correct code, while allowing legitimate calls to proceed to their destination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If IVR systems use confirmation schemes requiring callers to press buttons or provide data, then legitimate calls can be verified, but fraudsters using autodialers can circumvent this by emulating DTMF codes to mimic live callers
Solution Approach 1:
The patent replaces the mechanical DTMF code verification system with an acoustic analysis system that examines the actual sound waveforms of caller responses. Instead of relying on button presses that can be emulated, the system analyzes the acoustic properties of speech and natural DTMF tones generated by human callers, making it impossible for autodialers to circumvent since they cannot replicate human vocal characteristics.
Solution Approach 2:
The system changes the verification parameter from discrete DTMF frequency codes to continuous acoustic waveform characteristics. By analyzing parameters such as pitch, timbre, speech patterns, and temporal characteristics of the caller's voice, the system transforms the verification process into a dimension where autodialer emulation fails, while legitimate human callers continue to pass verification naturally.
2Reliability
If the system analyzes audio patterns of all incoming calls to detect fraud, then fraudulent calls can be accurately identified, but call processing time increases and may cause legitimate calls to be delayed
Solution Approach 1:
The system performs preliminary acoustic analysis during the initial ringing phase and first few seconds of call connection, before the call is fully established. By analyzing audio patterns in this early window, the system can identify fraudulent calls quickly and terminate them before significant processing time is consumed, while legitimate calls proceed normally after brief verification.
Solution Approach 2:
The system implements expedited processing for calls that exhibit clear fraudulent acoustic signatures during initial analysis. When fraud is detected in the early analysis phase, the system immediately terminates the call without proceeding through full verification sequences, thus minimizing time loss for fraudulent calls while maintaining accurate detection.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system is provided for thwarting the undesired incoming calls and eliminating the impact of robocalls, tollfree pumping, political spam, data collection bots, and other phone fraud forms. The system has a spam detection unit with a plurality of spam type modules and each spam type module is configured to detect a different type of spam.The incoming call is evaluated on a separate audio channel. If spam is detected an IVR prompts the caller to input a code for a second spam verification step.