Fig. 9From: Black-box adversarial attacks through speech distortion for speech emotion recognitionThree models after adversarial trainingBack to article page