Continuous Voice Biometrics for Multi-Speaker Profile Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice biometric systems face challenges in efficiently enrolling new speakers and maintaining accurate speaker profiles, particularly in dynamic environments with multiple speakers, leading to issues like staleness and false authentication.
Innovation Solution
A passive and continuous voice biometrics system that employs machine-learning models to extract embeddings from audio signals, allowing for flexible enrollment and profile management without explicit user prompts, and uses adaptive thresholding and unsupervised clustering to maintain accurate speaker identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If active enrollment is used to verify speaker identity, then authentication accuracy is improved, but user convenience deteriorates due to time-consuming enrollment processes
Solution Approach 1:
The system performs preliminary speaker verification in the background before actual authentication is needed. Speaker profiles are created and updated continuously from ambient audio captures, so when active enrollment is triggered, the verification process is already partially complete, reducing the time burden on users while maintaining accuracy
Solution Approach 2:
The system implements continuous speaker verification that operates in the background alongside active enrollment. Audio captures are continuously processed to update speaker profiles, ensuring that verification accuracy is maintained without requiring users to stop their natural interactions with the device
2Ease of manufacture
If static enrollment is used to initialize voice models, then device setup is simplified, but verification accuracy deteriorates over time as voices change
Solution Approach 1:
The system transitions from static enrollment to dynamic continuous verification. Speaker profiles are initially created during simple device setup, then continuously updated in the background as new audio captures are processed, allowing the system to adapt to voice changes over time while maintaining setup simplicity
Solution Approach 2:
The system performs self-updating of speaker profiles using ambient audio captures without requiring user intervention. The continuous verification process automatically detects and incorporates new speaker characteristics, maintaining accuracy without additional user effort or complex setup procedures
3Adaptability or versatility
If voice biometrics system processes multiple speakers concurrently, then multi-user support is improved, but speaker differentiation difficulty increases
Solution Approach 1:
The system segments the audio signal into individual speaker components using source separation techniques. Each speaker's audio stream is isolated and processed independently, allowing the system to maintain accurate speaker differentiation even when multiple users are present, thus supporting multi-user scenarios without increasing differentiation difficulty
Data Source
AI summary
Embodiments described herein provide for a voice biometrics system execute machine-learning architectures capable of passive, active, continuous, or static operations, or a combination thereof. Systems passively and/or continuously, in some cases in addition to actively and/or statically, enrolling speakers as the speakers speak into or around an edge device (e.g., car, television, radio, phone). The system identifies users on the fly without requiring a new speaker to mirror prompted utterances for reconfiguring operations. The system manages speaker profiles as speakers provide utterances to the system. Machine-learning architectures implement a passive and continuous voice biometrics system, possibly without knowledge of speaker identities. The system creates identities in an unsupervised manner, sometimes passively enrolling and recognizing known or unknown speakers. The system offers personalization and security across a wide range of applications, including media content for over-the-top services and IoT devices (e.g., personal assistants, vehicles), and call centers.