Deepfake Shield — Real-Time Liveness & Biometrics

Audio Verification

Speech spoofing detection via wav2vec2 self-supervised backbone

Drop voice clip here or browse

Supports .wav, .flac, .mp3 • Max 50 MB

speech_sample.wav

1.2 MB

Verified Human Confidence: 98.4%

Bonafide (Real) P(Spoof): 0.016 Synthetic (Spoof)

Processed Duration

4.00s (16 kHz)

Backbone Architecture

wav2vec2-base

Face manipulation detection via MTCNN + EfficientNet-B0

Drop video clip here or browse

Supports .mp4, .avi, .mov • Max 50 MB

face_clip.mp4

4.8 MB

Verified Human Confidence: 96.2%

Authentic (Real) P(Fake): 0.038 Manipulated (Deepfake)

Frames Analyzed

24 facial crops

Backbone Architecture

EfficientNet-B0