To the best of our knowledge, this is the first systematical study on generating adversarial audios for various acoustic systems, including speech.

Jan 23, 2019 · In this paper, we present SirenAttack, a new class of attacks to generate adversarial audios. Compared with existing attacks, SirenAttack highlights with a set ...

To the best of our knowledge, this is the first systematical study on generating adversarial audios for various acoustic systems, including speech.

Oct 5, 2020 · In this paper, we present SirenAttack, a new class of attacks to generate adversarial audios. Compared with existing attacks, SirenAttack ...

SirenAttack is evaluated on a set of state-of-the-art deep learning-based acoustic systems (including speech command recognition, speaker recognition and ...

In this poster, we present SIRENATTACK, a new class of adversarial attacks against deep neural network-based acoustic systems. Compared with prior work, ...

They concentrated on attacking the DNN model using optimization (opt), transfer (TA), interpolation, gradient estimation (GE), and evolutionary methods.

In this paper, we present SirenAttack, a new class of attacks to generate adversarial audios. Compared with existing attacks, SirenAttack highlights with a set ...

Adversarial Attacks Against Automatic Speech Recognition Systems via Psychoacoustic ... SirenAttack: Generating Adversarial Audio for End-to-End Acoustic Systems ...

@inproceedings{du2020sirenattack, title={Sirenattack: Generating adversarial audio for end-to-end acoustic systems}, author={Du, Tianyu and Ji, Shouling and ...