Investigate human voice removal from MV videos.

[YuantongL]

Investigate the feasibility of removing human voice from either video stream or downloaded video.

[YuantongL]

Tested this awesome repository https://github.com/tsurumeso/vocal-remover in python, it works great. Needs to figure out a way to integrate to our project.

[YuantongL]

After a few days investigation, both Vocal-remover on pytorch and Spleeter on TensorFlow works well. But it is a pain to integrate it with macOS app.
Feels like 2 steps is needed

1. Make a library that can split a piece of audio (I'm going to open up a new project to experiment this, possibly open up a new repo)
2. Figure out how to download Youtube video and apply the outcome of step 1

One other approach is to embed python in Xcode project as static library, and ship the whole thing. Downside is it is huge (a few hundred MBs), seems not acceptable.