Multi-Modal Multi-Task Deep Learning For Speaker And Emotion Recognition Of TV-Series Data | IEEE Conference Publication | IEEE Xplore