When doing audio and video sentiment recognition, I found that a lot of code is duplicated, often a function in different time debugging for a long time, based on this problem, I want to manage all the previous work, organized into an open source library can be iterative. For their own use and others.