Console Interface and Library to remove silent parts of a media file.
|Unedited (Before)||Processed by Unsilence (After)|
|Time before edit: 0:09:45 (100%)||Time after edit: 0:07:56 (81.2%), Difference: -0:01:50 (-18.8%)|
The MIT Intro at the beginning is not included into the time, since I left it in to show the license of the videos.
These videos are from this online lecture:
Ana Bell, Eric Grimson, and John Guttag. 6.0001 Introduction to Computer Science and Programming in Python. Fall 2016. Massachusetts Institute of Technology: MIT OpenCourseWare, https://ocw.mit.edu. License: Creative Commons BY-NC-SA.
More Information about Licensing can be found in the Licensing Segment of this README.
Unsilence is an open-source tool that removes silence from a media clip of your choice (audio, video).
You can use it to speed up videos without changing the audible speed, so you can understand everything, but get through a video faster.
Exemplary use cases
You are a college student and watch your lectures online (but have access to the video files). Instead of just increasing the playback speed to ~1.5x, you can remove
the parts that do not contain any value, like your lecturer thinking or waiting for something. But instead of cutting out these silent parts, speeding them up by a
different, much faster factor (think 6-8x) makes you still able to follow what is happening, so drawing or writing with no speech is sped up, which makes it far more pleasant to watch
You want a video editor that automatically cuts any time you talk (or make any sound). That could be useful for manual time lapses
(you make a sound every time a short segment should be recorded), or for very fast jump cut videos with no manual editing required
You want to have some fun and remove all the audible parts from a video, leaving only the parts where nearly silent noises are in the video (breathing, writing, ...)
Unsilence can be used as a console line interface or as a python library, with which you can develop your own projects
- Python >= 3.8.0
- pip (should be installed automatically with python, could be different on some linux distros)
- ffmpeg >= 4.2.0
Installation as command line interface (using pip and pipx)
# Installing pipx pip install pipx # Installing Unsilence as Command Line Software pipx install unsilence # If pipx asks you to, you also need to execute the following line # as well as close and reopen your terminal window pipx ensurepath
Installation as library (using pip)
# Installing Unsilence as Command Line Software pip install unsilence
Installation as command line interface (from source)
# Clone the repository (stable branch) git clone -b master https://github.com/lagmoellertim/unsilence.git unsilence #Change Directory cd unsilence # Install pip packages pip install -r requirements.txt pip install pipx # Install unsilence package pipx install .
Installation as library (from source)
# Clone the repository (stable branch) git clone -b master https://github.com/lagmoellertim/unsilence.git unsilence #Change Directory cd unsilence # Install pip packages pip install -r requirements.txt # Install unsilence package python3 setup.py install
Basic Command Line Usage
This generates a new file, where the silent parts are 6x as fast as before, the audible parts are the same speed as before
unsilence [input_file] [output_file]
You can change the speed of audible parts with
-as [speed], the speed of silent parts with
unsilence [input_file] [output_file] -as [speed] -ss [speed]
You can change the volume of audible parts with
-av [volume], the volume of silent parts with
unsilence [input_file] [output_file] -av [volume] -sv [volume]
To generate an audio only output file, you can add the
unsilence [input_file] [output_file] -ao
To speed up the rendering process, you can increase the thread count using
unsilence [input_file] [output_file] -t [threads]
For many more settings, type
Basic Library Usage
Take a look at this example
For this project, I took inspiration from the CaryKH's video Jumpcutter.
This project does not share any source code with his implementation, and is more optimized for my use case (fast and efficient lecture silence removal).