On the Exploitability of Audio Machine Learning Pipelines to Surreptitious Adversarial Examples

Open in new window