Prepend Acoustic Attack
A public research-code contribution exploring universal audio segments that can mute Whisper speech recognition.
Open the original ↗This is Shmulik’s public contribution and experiment snapshot based on the upstream Muting Whisper codebase. The project investigates a short universal acoustic segment that can cause Whisper-style speech recognition to ignore incoming speech.
My fork records additional experiments, results, scripts, and cleanup. The underlying Muting Whisper research and implementation are credited to the upstream authors.