Fig. 2From: Towards the transferable audio adversarial attack via ensemble methodsTransferable attacks on speech recognition modelsBack to article page