VoiceLive - A Phoneme Localization Based Liveness Detection for Voice Authentication on Smartphones

Explore a conference talk on VoiceLive, an innovative phoneme localization-based liveness detection system for voice authentication on smartphones. Delve into the challenges of voice spoofing attacks and discover how this novel approach leverages human speech system characteristics and phoneme segmentation to enhance security. Learn about the system's architecture, attack models, and experimental methodology, including data collection and attack design. Examine the overall performance and robustness of VoiceLive against phone displacement, gaining insights into cutting-edge mobile security solutions presented at the 23rd ACM Conference on Computer and Communications Security.

Intro
Mobile Applications
Voice Authentication: Spoofing
Existing Solutions
VoiceLive: Basic Idea
System & Attack Models
Human Speech System
Phonemes: Vowels
Phoneme Localization with Mic Array
Approach Overview
System Overview
Phoneme Segmentation
TDOA Calculation
Similarity Comparison Normalize TDA dynamic to the same scale as the template
Properties Of TDOA dynamics
Experiment Methodology
Data Collection
Attack Design
Overall Performance
Robustness to Phone Displacement
Conclusion