VoiceBlock: Privacy through Real-Time Adversarial Attacks with Audio-to-Audio Models

INIS

• perturbations 100%
• speech 100%
• data 50%
• learning 50%
• security 50%
• architecture 50%
• implementation 50%
• optimization 50%
• filters 50%
• streams 50%
• neural networks 50%
• impulse 50%

Computer Science

• Models 100%
• Adversarial Machine Learning 100%
• User 75%
• Speaker Recognition 50%
• Neural Network Model 25%
• Anonymization 25%
• Learning Systems 25%
• Recognition System 25%
• Security and Privacy 25%
• Finite Impulse Response Filter 25%
• Deep Learning 25%
• Contexts 25%
• Thread 25%
• Speech Enhancement 25%
• de-noising 25%
• Adversarial Example 25%