Defensive-Publications

Defensive Publication · Prior Art Record
Novel Authentication Method Prevents AI Voice Cloning Through Neural–Acoustic Correlation
Multi-modal biometric verification using EEG↔speech correlation to defeat synthetic voice and replay attacks.
Document ID: GABEY-DP-2026-02-06-MULTIMODAL-VOICE-VERIFICATION-v1.0Version: v1.0
Download PDF Download Markdown Download SHA-256 Hashes Download OpenTimestamps Proof (.ots)
Development commenced (claimed)
3 Feb 2026 (Australia/Melbourne, AEDT, UTC+11)
Architecture specification completed
5 Feb 2026 (Australia/Melbourne, AEDT, UTC+11)
Public disclosure published (verifiable)
6 Feb 2026, 10:55 PM (Australia/Melbourne, AEDT, UTC+11) [= 6 Feb 2026, 11:55 UTC]
SHA-256 (PDF)
cbb7164536b1a173da3e6a48f8585a64c48641b6696bd2fc4d78ddefeee9e9f2
SHA-256 (MD)
e58400d622e4411027a758975cb0c9a9e2d6ad570262be4f169e8f5bfc7245dc
Contact
gabeyinfo [at] gmail [dot] com Contact / Support
Inventor contact is recorded in the publication artefacts (PDF/MD).
Notice
This disclosure is a defensive publication intended to establish prior art. Not a patent application.

Abstract

This defensive publication discloses a multi-modal biometric authentication system that resists AI voice cloning, deepfake audio, and replay attacks by verifying a neural–acoustic correlation between EEG activity and speech production. Conventional voice biometrics rely primarily on acoustic characteristics and can be defeated by modern speech synthesis and conversion. The disclosed method introduces a physiological verification layer that confirms speaker presence and cognitive intent by assessing timing and statistical alignment between EEG features and phoneme-level speech events.

Disclosure Summary (non-limiting)

  • Captures EEG and speech audio during enrolment and authentication.
  • Aligns EEG features to speech events using a synchronisation layer (offset/drift handling).
  • Computes a neural–acoustic correlation score (e.g., cross-correlation/coherence/mutual information/sequence alignment).
  • Applies policy gating (voice match + correlation threshold) and records an audit/evidence outcome.

Integrity & Verification

Note: v1.1 will expand this disclosure (e.g., figures/diagrams and additional technical detail) while v1.0 remains the initial disclosure record.

DigitalPlank Hub

AI, Cybersecurity & Electronics

© 2025 GABEY Consulting Pty Ltd

FAQs on Design & Development

NOMATEQ Defensive Publications — verifiable prior art records for security innovation.
View publications Contact / Support
What’s the difference between a defensive publication and a patent?

A patent (where granted) can provide enforceable exclusive rights for a limited term. A defensive publication is a public disclosure intended to establish prior art, reducing the chance others can later patent the same idea.

Side-by-side
Aspect Patent Defensive publication
Primary goal Exclusive rights / licensing leverage Prevent others from patenting (prior art)
Timeline Often months–years (jurisdiction dependent) Immediate on publication
Outcome Enforceable rights (where granted) Public record for prior art and verification
Who can implement? Typically patent holder/licensees only Concepts are publicly disclosed (see caveats below)
Note: A defensive publication is not legal advice and does not itself grant a “license.” If you’re commercialising at scale, consider an IP review for your specific implementation and jurisdiction.

Why NOMATEQ uses defensive publications

  • Establish prior art for security concepts so they can’t be locked up later.
  • Enable adoption by reducing patent uncertainty around the disclosed ideas.
  • Provide integrity evidence (hashes, timestamps, archive captures) for audit and provenance.
Can I implement these concepts in a commercial product?

Quick answer

Generally, yes - defensive publications are intended to make concepts publicly known and usable. However, you should still consider the broader IP landscape for your specific implementation.

 ​Typical use cases

  • Implement in commercial or non-commercial products
  • Use in proprietary or open-source systems
  • Modify and extend the ideas for your environment
   Practical caveats (worth reading)
  • No warranty: published “as is” - validate for your threat model and environment.
  • Compliance: you remain responsible for privacy/security obligations and sector requirements.
  • IP checks: your implementation could still intersect with third-party patents or constraints. For high-value deployments, do an IP review.

If you want clarification on scope or implementation considerations, use the  Contact / Support form and include the Document ID.

How do I verify publication integrity (PDF/Markdown/Hashes)?

Quick answer

Each publication includes SHA-256 hashes so you can independently verify that the artefacts you downloaded match the published record.

Recommended quick-check (Linux/macOS)

sha256sum -c GABEY-DP-2026-02-06-MULTIMODAL-VOICE-VERIFICATION-v1.0.HASHES.txt

Manual verification
Windows (PowerShell):

Get-FileHash -Algorithm SHA256 .\filename.pdf

macOS/Linux (Terminal):

shasum -a 256 filename.pdf

  Match: file is authentic and unmodified

  Mismatch: file differs - re-download from the official page and re-check

Extra integrity evidence you may see referenced
  • OpenTimestamps proof (.ots): timestamp evidence for the hashes file
  • Wayback captures: archive snapshots for the page and artefacts
  • Multi-source consistency: hashes + timestamp + archive alignment
Tip: When requesting help, include your OS, the filename, and whether you verified the HASHES.txt file or computed hashes manually.
More questions?

For questions not covered here, use the contact form:

Contact / Support

NOMATEQ: Building security through transparency.

Tip: Include the Document ID and which artefact (PDF/MD/HASHES) you’re referring to.