Learning deep multimodal affective features for spontaneous speech emotion recognition 张石清, 陶欣, 闯跃龙, 赵小明 3月 1, 2021 DOI