Voice Call Watermark Authentication Against Audio Manipulation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing voice call authentication methods are vulnerable to fraud, where criminals record and manipulate voice data to impersonate individuals for fraudulent transactions.

Innovation Solution

A method and system that generates a call-specific watermark pattern in voice calls, allowing authentication by detecting the number of watermarks in the audio stream to verify its authenticity, using a network node and verification server.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If voice recognition mechanisms are used to confirm contractual relationships, then contract establishment efficiency is improved, but security against voice recording and manipulation attacks deteriorates

Engineering Contradiction:
Improvecontract establishment efficiencyVSAvoidsecurity against voice manipulation
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system embeds watermarks in the audio stream during the voice call connection, before the voice data can be recorded and manipulated for fraudulent purposes. This preliminary action of watermarking the audio stream allows later detection of whether the voice data originates from an authentic call, preventing unauthorized contract establishment

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The watermark acts as an intermediary marker embedded in the audio stream that carries authentication information. This watermark intermediary enables the system to verify the authenticity of voice data without disrupting the normal voice recognition contract establishment process

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If audio streams are analyzed for watermark detection, then authenticity verification is improved, but processing complexity increases

Engineering Contradiction:
Improveauthenticity verification accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system extracts only the watermark patterns from the audio stream for authentication purposes, rather than analyzing the entire audio content. This extraction approach maintains high authenticity verification accuracy while significantly reducing processing complexity compared to full audio analysis

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The watermark detection process changes the analysis parameter from examining voice content characteristics to detecting specific embedded watermark patterns. This parameter change simplifies the detection process while maintaining accurate authenticity verification

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3582465B1Solution for determining an authenticity of an audio stream of a voice call
Publication Date: 2026.04.29 TELIASONERA AB
  • EP3582465B1 patent drawingFigure 1A~1B
  • EP3582465B1 patent drawingFigure 2~4
  • EP3582465B1 patent drawingFigure 3

AI summary

The present invention relates to a method for determining an authenticity of at least a portion of an audio stream of a voice call connection. In the method it is received a request (410) for confirming an authenticity of at least portion of an audio stream of a voice call connection; detected a number of voice call connection specific watermark patterns included in the portion of the audio stream in question; and set (430), in accordance with a detection of the number of the voice call connection specific watermark patterns included in the portion of the audio stream, a detection result to express one of the following: i) the portion of the audio stream is authentic, ii) the portion of the audio stream is nonauthentic. The invention also relates to a network node (210), a computer program product and a system.