Cassette Bot

🤖 📼 Commandline tool for remixing videos with timecoded transciptions.

Requirements

ffmpeg (A complete, cross-platform solution to record, convert and stream audio and video)

$ brew install ffmpeg

Installation

make build

Or to build with examples:

make build_with_examples

Workflows

Speech

Produce a video spliced to match the input_text.

$ cassettebot speech \
    examples/mark/video.mp4 \
    examples/mark/transcription.json \
    --input_text='i wanna talk about this specific situation when i was in college and I started facebook' \
    --video_output='examples/mark/results/speech.mp4' \
    --verbose

See result: examples/results/speech.mp4

Silence

Produce a video from the original with all speech instances removed.

$ cassettebot silence \
    examples/mark/video.mp4 \
    examples/mark/transcription.json \
    --video_output='examples/mark/results/silence.mp4'

See result: examples/results/silence.mp4

General Use

usage: Cassette Bot [-h] [--input_text INPUT_TEXT]
                    [--video_output VIDEO_OUTPUT] [--verbose]
                    {speech,silence,markov} video_source transcript

positional arguments:
  {speech,silence,markov}
                        type of product: {'speech', 'silence', 'markov'}
  video_source          path of video to process
  transcript            path of timecoded json transcript

optional arguments:
  -h, --help            show this help message and exit
  --input_text INPUT_TEXT, -i INPUT_TEXT
                        input text
  --video_output VIDEO_OUTPUT, -o VIDEO_OUTPUT
                        output directory path
  --verbose, -v         check created transcript before building video

🤖 📼 Pull requests welcome

Research notes

• https://github.com/antiboredom/videogrep
• https://www.clips.uantwerpen.be/pages/pattern-search
• https://github.com/bbc/kaldiboy — saw after making this.