Fig. 8From: Black-box adversarial attacks through speech distortion for speech emotion recognitionDescription of the three models under the modulation spectrum smoothing attackBack to article page