.d8888b.  888 d8b                   d8b 
d88P  Y88b 888 Y8P                   Y8P 
888    888 888                           
888        888 888 88888b.  88888b.  888 
888        888 888 888 "88b 888 "88b 888 
888    888 888 888 888  888 888  888 888 
Y88b  d88P 888 888 888 d88P 888 d88P 888 
 "Y8888P"  888 888 88888P"  88888P"  888 
                   888      888          
                   888      888          
                   888      888          

A tool for capturing moments of interest

Overview

Clippi is a video editing and production tool that aims to capture moments of interest in long form content (podcasts, streams etc.)

Features

Capture moments of interest

Clippi leverages ML pipelines and media processing algorithms to analyze text, audio, and visual data. The metadata is used to produce "weight embeddings" on sections of the video based on perceived interest/desire. The moments of interest are delivered as timestamp indices representing the best clips found.

Analyze metadata to deliver in-depth content analysis

With clips that are of interest, Clippi can generate natural language varying in depth depending on the use case. (descriptions, captions, chapters, summaries etc.)

Architecture

Usage

install from source

λ git clone https://github.com/cameronslee/clippi.git
λ cd clippi

install packages

λ pip install -r requirements.txt

run

λ python3 /clippi/clippi.py 

Docker

λ docker build -t clippi . 
λ docker run -dit clippi 
λ docker ps -a 

# can attach to the session with the following:
λ docker exec -it <container_name> bash

Final Notes

This project is still in development.