EarDynamic: An Ear Canal Deformation Based Continuous User Authentication Using In-Ear Wearables
summary
A continuous, passive authentication system uses an in‑ear wearable’s built‑in speaker and microphone to emit inaudible beeps and record reflected signals, extracting channel responses that capture both static ear canal geometry and dynamic deformations tied to speech. Phonemes are grouped into 7 phoneme categories to map articulatory gestures to deformation features, fed into a classifier. A prototype runs on earbuds, reports recall 97.38% and F1 96.84% (overall ≈93% accuracy) and tolerates noisy daily activities.
AI-written summary · about
chapters (5)
- 0:00Start
- 0:21Background Information
- 2:31Static Geometry
- 3:25How Do We Measure Dynamic after that the Ear Canal Deformation
- 4:38Static and Dynamic Scenarios
related talks
12mSkullConduct: Biometric User Identification on Eyewear Computers Using Bone Conduction Through ...
06mVocalLock: Sensing Vocal Tract for Passphrase-Independent User Authentication Leveraging ...
33mFor Smarter Authentication, We Might Need to Use ...
06mListen to Your Fingers: User Authentication based on Geometry Biometrics of Touch Gesture
16mEchoLLM: LLM-Augmented Acoustic Eavesdropping Attack on Bone Conduction...
35mThe UX Of Advanced Voice Biometrics For Authentication
33mYour voice confirms my identity
19mThumprint: Socially-Inclusive Local Group Authentication Through Shared Secret Knocks