K
KITT
Text
Continuous Biometric Authentication via Lip Articulator Dynamics
A biometric authentication system that identifies a person from the dynamic motion of their lip articulators during speech rather than from static lip shape, giving shape-independent and continuous verification. Lip motion is represented as a hierarchy of articulatory-dynamic features and matched with a Siamese neural network combined with an ensemble classifier. Using only lip movement, it avoids collecting full facial data and resists mimicry and AI-deepfake spoofing attacks.