Automatic Video Editing

Automatic video editing with Python, moviepy and vosk API

Tutorial

You can read detailed tutorial on Towards Data Science.

Results

See the videos:

• Automatic Video Editing - Control Words on Youtube
• Automatic Video Editing - Silence on Youtube

Setup

First, you need to install moviepy and vosk libraries. You can do it with pip install -r requirements or pip install moviepy vosk command.

Next, you need to download the vosk model from official site and unzip it.

Next open automatic_video_cutter.py file with any text editor and set the parameters before calling a main method (in 305 line):

• model_path - path to unziped vosk model;

• video_path - path to video file to convert;

• result_path - new filename to save final video;

• silence - processing method. If True, process video with silence mode, if False - with control words;

• bitrate - bitrate for write_videofile function. Default is None, must be like '2500k', '5000k', '10000k' etc.

• threshold - threshold of silence time in seconds. Used only if silence==True;

• offset_silence - offset in seconds. Used only if silence==True;

• start_word - control word that signals the beginning of the video fragment to be cut. Used only if silence==False;

• end_word - control word that signals the ending of the video fragment to be cut. Used only if silence==False;

• offset_words - offset in seconds. Used only if silence==False.

Now you can run the script using python automatic_video_cutter.py command.

If you can use Jupyter Notebook you can open automatic_video_cutter.ipynb file and specify these parameters in Set parameters 14 cell. Path to the vosk model is specified earlier - in Loading a vosk model section, 11 cell.

Project Structure

The project has the following structure:

automatic_video_editing
├───models
│   ├───vosk-model-en-us-0.21
│   └───vosk-model-ru-0.10
├───videos
│   .gitignore
│   
│   automatic_video_cutter.ipynb
│   automatic_video_cutter.py
│   Word.py
│
│   README.md
└── requirements.txt

It contains the following folders:

• The models folder contains vosk models downloaded from official site.
• The videos folder contains videos to be processed.

And code files:

• Word.py file describes the Word class.
• automatic_video_cutter.ipynb notebook contains a full program.
• automatic_video_cutter.py file contains the same full program but in the form of a python script. It can be used to run a program without Jupyter Notebook.