English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper we investigate speech denoising as a defense againstadversarial attacks on automatic speech recognition (ASR) systems. Adversarialattacks attempt to force misclassification by adding small perturbations to theoriginal speech signal. We propose to counteract this by employing aneural-network based denoiser as a pre-processor in the ASR pipeline. Thedenoiser is independent of the downstream ASR model, and thus can be rapidlydeployed in existing systems. We found that training the denoisier using aperceptually motivated loss function resulted in increased adversarialrobustness without compromising ASR performance on benign samples. Our defensewas evaluated (as a part of the DARPA GARD program) on the 'Kenansville' attackstrategy across a range of attack strengths and speech samples. An averageimprovement in Word Error Rate (WER) of about 7.7% was observed over theundefended model at 20 dB signal-to-noise-ratio (SNR) attack strength.

Perceptual-based deep-learning denoiser as a defense against adversarial attacks on ASR systems