Audio-Driven Facial Landmark Generation in Violin Performance using 3DCNN Network with Self Attention Model | IEEE Conference Publication | IEEE Xplore