Fig. 7From: End-to-end speech emotion recognition using a novel context-stacking dilated convolution neural networkComparisons between using raw time domain samples and using log mel-spectrograms as the SER model inputBack to article page