Audio-Visual Fusion With Temporal Convolutional Attention Network for Speech Separation | IEEE Journals & Magazine | IEEE Xplore