Zero-Knowledge Voiceprint Sharing via Encrypted Comparison Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional voiceprint sharing technologies fail to provide secure, zero-knowledge sharing and comparison of voiceprints, leading to vulnerabilities in authentication systems as plaintext voiceprints can be accessed maliciously during storage and transit, compromising privacy across multiple communicating parties.
Innovation Solution
Implementing a system that allows encrypted voiceprints to be shared and compared without revealing the unencrypted voiceprints, using cryptographic protocols like Diffie-Hellman key exchange and elliptical curve cryptography to calculate similarity scores between encrypted voiceprints, ensuring each party only accesses its own plaintext voiceprints and maintains privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If plaintext voiceprints are stored and transmitted for comparison, then authentication accuracy is improved, but security and privacy are compromised due to malicious access during storage and transit
Solution Approach 1:
The patent introduces encrypted voiceprint models as an intermediary representation. Instead of directly sharing and comparing plaintext voiceprints, the system uses encrypted models that can be mathematically compared without revealing the underlying voiceprint data. This mediator enables authentication functionality while maintaining security throughout storage and transmission.
Solution Approach 2:
The patent transforms voiceprints from plaintext format to encrypted model format, changing the parameter state of the data. The encrypted voiceprint models contain the necessary information for comparison and authentication but exist in a transformed state that prevents direct malicious access to the original voiceprint information during storage and transit.
2Object-affected harmful factors
If encrypted voiceprint models are used for sharing, then security is improved, but the ability to perform comparison operations deteriorates without access to plaintext
Solution Approach 1:
The patent replaces the mechanical approach of decrypting voiceprints for comparison with a cryptographic mathematical system. Encrypted voiceprint models can be directly compared using encrypted comparison operations and similarity scoring functions that work on encrypted data, eliminating the need to mechanically decrypt the voiceprints to perform comparisons.
Solution Approach 2:
The patent changes the operational parameters by implementing cryptographic comparison operations that work directly on encrypted data. Instead of requiring plaintext for comparison, the system uses encrypted comparison models and similarity scores that operate in the encrypted domain, maintaining both security and comparison capability.
3Adaptability or versatility
If multiple parties share voiceprints for authentication, then authentication functionality is improved, but privacy preservation deteriorates as parties gain knowledge of each other's voiceprints
Solution Approach 1:
The patent uses encrypted voiceprint models as intermediaries that enable multiple parties to perform authentication operations without directly exposing their actual voiceprints. Each party can contribute to the authentication process using their encrypted models, and the system can compute similarity scores and verify identities without any party gaining knowledge of the other parties' plaintext voiceprints.
Solution Approach 2:
The patent segments the authentication process into separate cryptographic operations that can be performed independently on encrypted data. Each party's voiceprint remains segmented and encrypted throughout the process, with only the necessary cryptographic computations performed on the encrypted representations, preventing information leakage while maintaining authentication functionality.
Data Source
AI summary
Disclosed herein are embodiments of systems and methods for zero-knowledge multiparty secure sharing of voiceprints. In an embodiment, an illustrative computer may receive, through a remote server, a plurality of encrypted voiceprints. When the computer receives an incoming call, the computer may generate a plaintext i-vector of the incoming call. Using the plaintext i-vector and the encrypted voiceprints, the computer may generate one or more encrypted comparison models. The remote server may decrypt the encrypted comparison model to generate similarity scores between the plaintext i-vector and the plurality of encrypted voiceprints.