Robust Adversarial Muting Attacks on Whisper

Devised an embedded adversarial audio attack on OpenAI Whisper that broke 95% of transcription tasks.

This research explores the intersection of speech technologies, adversarial machine learning, and model robustness.