Attacking Speaker Recognition With Deep Generative Models

Abstract

In this paper we investigate the ability of generative adversarial networks (GANs) to synthesize spoofing attacks on modern speaker recognition systems. We first show that samples generated with SampleRNN and WaveNet are unable to fool a CNN-based speaker recognition system. We propose a modification of the Wasserstein GAN objective function to make use of data that is real but not from the class being learned. Our semi-supervised learning method is able to perform both targeted and untargeted attacks, raising questions related to security in speaker authentication systems.

5 pages, 3 Figures, 1 table

References in corpus (5)

• Adam: A Method for Stochastic Optimization
• Conditional Generative Adversarial Nets
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Improved Training of Wasserstein GANs
• Tacotron: Towards End-to-End Speech Synthesis

Cited by in corpus (5)

• Preech: A System for Privacy-Preserving Speech Transcription
• TequilaGAN: How to easily identify GAN samples
• Learning to navigate image manifolds induced by generative adversarial networks for unsupervised video generation
• Spoofing Speaker Verification Systems with Deep Multi-speaker Text-to-speech Synthesis
• Neural voice cloning with a few low-quality samples