Fig. 1From: Towards the transferable audio adversarial attack via ensemble methodsCompare the frequency domain features after the inclusion of silence framesBack to article page