Deep Feed-Forward Sequential Memory Networks for Speech Synthesis | IEEE Conference Publication | IEEE Xplore