Parallelized-Video-Remove-Silence-Call

This Batch file folder can be used to execute Video-Remove-Silence from https://github.com/excitoon/video-remove-silence on shorter chunks of a video, in order to decrease the computation time using a simple form parallelization. This is especially interesting, as all the optimization is achieved outside of modifying the code of video-remove-silence, and instead decreasing the video length and using multiple instances of CMD to solve this issue. To run it, python is needed, as well as ffmpeg, which executes all the split- and merge commmands.

How it operates

1. Call.bat is executed, either by direct access or via a shortcut. This will go to the Desktop, and merge the audio of a.mp4 to the video of v.mp4 and thus create in.mp4. This is useful, if for example, a Lecture has two perspectives, one on the Lecturer and one on the Presentation. Then a.mp4 and v.mp4 are deleted, and in.mp4 is moved to Shorten\Files. If there is only one video file, it can be named in.mp4 and no a.mp4 and v.mp4 are needed.

2. MasterStart.bat is called, it will execute split.bat, which in turn splits the video in.mp4 into 7 minute long segments inside of Shorten\Files, but that length can of course be changed.

3. All instances of start000.bat etc. are called, they will each work on a segment created previously, and MasterStart.bat will wait for all of these processes to finish. The number of 19 .bat files is chosen arbitrarily, and can of course be adapted if necessary, especially if the source video file is too large, then two options would be, to either increase the segment size, or to add start019.bat etc.

4. Upon Completion, merge.bat is called from inside MasterStart.bat, which will reassemble all the segments into one single video file, named shortened.mp4 and it will also delete all the temporary video files created previously.

5. Call.bat will then move shortened.mp4 to the desktop, as well as the (merged) in.mp4 in case the user still needs this file.

6. Loudnorm.bat can be used to initially normalize a video's volume, as well as resample it to 41000hz in order to prevent any issues when feeding it into wave.py during the silence-removal process