Dialogueify Video

A small CLI python utility created for cutting down on non-dialogue sections of video utilising a subtitle file.

Prerequisites

• Python 3+
• ffmpeg version n5.1+ (will probably work with older versions too, not tested)
• Twice the amount of disk space as the output (this is because the program uses the system's tmp folder to store the clips with dialogue)
• Terminal which supports ANSI/VT100 escape sequences

Usage

0. Clone this repo
1. Open command prompt, powershell, or other terminal software
2. Open the .py file with python along with the needed arguments: py dialogueify.py <video.mp4> <subtitles.xxx> <Output Filename> [-ao]

Arguments and flags

• <video.mp4> - Path to video file - Format: forward slashes, example: c:/users/dummy/desktop/video.mp4
• <subtitles.xxx> - Path to the subtitle file - Format: forward slashes, example: c:/users/dummy/desktop/subs.srt
  • Currently, formats .srt and .json3 are supported for subtitle files, and all video formats supported by ffmpeg for video files.
• <Output filename> - Name of output file - The file will be given an extension based on the flags. The default is .mp4, example: ./dielogueified_video.mp4
• [-ao] - Audio Only - Generates a .mp3 file instead of the default .mp4, much faster than the default

Implementation

The program first analyzes the video based on the timing of the subtitles. By defaut, the minimum length of each clip is (clip length + 1 second). By default, clips will be grouped together and kept in one piece if there is a 3 second or smaller delay inbetween two subtitles, this is in place to keep together continuity with subtitles which are close together. After each clip/group is analyzed, a temporary clip is saved in the system's temporary storage After the whole length of the video is analyzed, the clips are stiched together, and output to the given filename

Mission

This script was inspired by this language learning video, and specifically for the French version of Spider Man: Into The Spiderverse, which was reduced from an original runtime of 1:56:xx to a new runtime of 1:17:xx. Cutting down on non-dialogue parts of videos allows language learners to focus on speech nstead of having to see all of a video/movie.

Roadmap

• More customization
  • More output formats
  • Customizable timing for padding and grouping
    • Account for padding being bigger than grouping
• Allow pairing of audio files with subtitles
• Long term: create a swiss army knife of utilities and include this one in it
• More input validation of arguments and proper handling of improper input
• Descriptive error messages with error handling (improper suttiles format, insufficient space...)
• Fix double big-progress bar bug