transform_video.py doesn't duplicate audio about fast-style-transfer

[AndyZhu1991]

I have created a pull request to solve this issue:
#76

[julien-c]

It should actually style transfer it, not duplicate :)

[Schwusch]

This could possibly be fixed with piping the video along with the audio from ffmpeg into the Python script, and piping the styled video along with the original audio into another ffmpeg instance.

This could also be solvable running ffmpeg afterwards combining the new video with the old audio as in:
http://stackoverflow.com/questions/11779490/how-to-add-a-new-audio-not-mixing-into-a-video-using-ffmpeg

I made a pull request with an example of piping video only, but I can see if audio is doable.

[lengstrom]

@Schwusch That sounds great! Pull requests are welcome :)

[xiesiyuan]

by using moviepy , maybe easily add audio.

[xiesiyuan]

import moviepy