Deepfake Detection

Compares audio waveforms with lip movement patterns to detect mismatches common in face-swap deepfakes.

Monitors chest and shoulder micro-movements for natural breathing rhythm that deepfakes often fail to replicate.

Detects subtle involuntary facial movements that last fractions of a second, a key authenticity signal.

Analyzes blink frequency and duration against natural baselines, as early deepfakes rarely blink.

Examines skin texture consistency for synthetic smoothing artifacts and unnatural boundary transitions.

Evaluates 3D head rotation consistency, checking for impossible or unnatural head orientations.