Real-time, Robust and Adaptive Universal Adversarial Attacks Against Speaker Recognition Systems

Keyphrases

• Speaker Recognition System 100%
• Universal Adversarial Attack 100%
• Neural Network 40%
• Deep Neural Network 40%
• Adversary 20%
• Popular 20%
• Audio 20%
• White-box 20%
• Generation Time 20%
• Adaptive Adjustment 20%
• Public Dataset 20%
• Attack Strategy 20%
• D-vector 20%
• Utterance 20%
• Over the Air 20%
• Agnostic 20%
• Attack Success Rate 20%
• Voice User Interface 20%
• Attack Generation 20%
• Adversarial Perturbation 20%
• Room Impulse Response 20%
• Universal Attack 20%
• Native English Speakers 20%
• Universal Perturbation 20%
• X-vector 20%
• Target Labels 20%

Computer Science

• Adversarial Machine Learning 100%
• Recognition System 100%
• Speaker Recognition 100%
• Deep Neural Network 40%
• User Interfaces 20%
• Generation Time 20%
• Imperceptibility 20%
• Impulse Response 20%
• English Speaker 20%
• Attack Generation 20%