Voiceprint CAPTCHA for Human-Machine Authentication

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current CAPTCHA techniques are vulnerable to attacks by machine learning algorithms and result in poor user experiences due to their susceptibility to cracking and interference with human recognition.

Innovation Solution

A voice-based CAPTCHA system that establishes a voiceprint for users during registration and logon, using dynamic verification codes and voiceprint recognition to distinguish humans from machines, combined with traditional logon credentials for enhanced security and usability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional CAPTCHA techniques (text, image, sound) are used, then machine attacks are partially prevented, but they are vulnerable to cracking by machine learning algorithms and result in poor user experiences

Engineering Contradiction:
Improvesecurity against machine attacksVSAvoiduser experience
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The patent changes the fundamental parameter of CAPTCHA from visual/audio recognition to voiceprint biometric recognition. By transforming the authentication mechanism from challenging users with puzzles to verifying unique vocal characteristics, it simultaneously improves security (machines cannot replicate human voiceprints) and user experience (natural voice-based authentication is more convenient than solving CAPTCHA challenges)

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical system of visual inspection and puzzle-solving with acoustic field-based voiceprint recognition. Instead of requiring users to visually distinguish distorted text or images, the system uses voice signal processing and biometric verification, substituting one interaction modality (visual-manual) with another (acoustic-automated) that is more resistant to automation attacks

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Object-affected harmful factors

If text CAPTCHA with distorted pictures and interference pixels is used, then machine recognition is prevented, but human recognition becomes difficult and user experience deteriorates

Engineering Contradiction:
Improvemachine recognition capabilityVSAvoidhuman recognition ease
Core Design Contradiction:
Object-affected harmful factorsVSEase of operation

Solution Approach 1:

The patent converts the inherent vulnerability of visual CAPTCHA (where distortion makes recognition difficult for both humans and machines) into a benefit by switching to voiceprint recognition. The 'harm' of making visual elements unrecognizable is transformed into the 'benefit' of using a modality (voice) that is naturally difficult for machines to synthesize and replicate, while remaining easy for humans to provide

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

Solution Approach 2:

Instead of making visual elements harder to recognize to prevent machine cracking, the patent inverts the approach by using a completely different modality (voice) where the natural characteristics of the medium itself provide security. Rather than obscuring the authentication element, it uses the unique biological characteristics of human voice production as the security mechanism

Inventive Principle:
Principle #13The other way round (Inversion)

3Reliability

If image CAPTCHA with complex classification and orientation recognition is used, then machine cracking is resisted, but the system requires extensive database support and is difficult to produce at scale

Engineering Contradiction:
Improveresistance to machine crackingVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the core security function from complex image databases and visual puzzle systems, isolating it into a voiceprint biometric verification system. By removing the need for extensive image databases, orientation recognition algorithms, and complex visual challenge generation, the system achieves high security with reduced complexity in data storage and processing requirements

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP2965249B1Method and system for distinguishing humans from machines
Publication Date: 2019.11.06 ALIBABA GROUP HOLDING LTD
  • EP2965249B1 patent drawingFigure 1
  • EP2965249B1 patent drawingFigure 2
  • EP2965249B1 patent drawingFigure 3~4

AI summary

A method and an apparatus for distinguishing humans from computers. During user registration, a computer prompts a human user to provide a spoken response to certain authentication information for registration. The computer obtains registration voice data from the spoken response and establishes a registration voiceprint of the human user. During user logon, the computer identifies the user requesting to logon by the user's logon credentials, provides authentication information for logon to the user, and prompts the user to provide a spoken response to the authentication information for logon. The computer obtains logon voice data from the spoken response, and establishes a logon voiceprint of the user. The computer then determines whether the user requesting to logon is human by comparing the logon voiceprint with the registration voiceprint.