Created by: daming-lu
After talking to Yibing from DeepSpeech team, it would be better if VisualDL could support 3-D real-time rendering like below:
