video-summary-a3b2c1 0.1.0

A pipeline for extracting summaries from videos using audio and visual context.

Video Summary

A Python package for extracting comprehensive summaries from videos using both audio (transcription) and visual (keyframe description) context.

Features

• Audio Transcription: Uses faster-whisper for efficient and accurate transcription.
• Visual Context: Extracts keyframes from important segments and describes them using GPT-4o Vision.
• Intelligent Synthesis: Combines audio and visual descriptions into a final, coherent summary.
• Modular Design: Easy to integrate into any Python project.

Installation

pip install video_summary

Note: Requires ffmpeg to be installed on your system.

Quick Start

import asyncio
from video_summary import VideoSummaryPipeline

async def main():
    # Initialize the pipeline
    pipeline = VideoSummaryPipeline()

    # Run on a video file
    summary = await pipeline.run("path/to/your/video.mp4")

    print("Video Summary:")
    print(summary)

if __name__ == "__main__":
    asyncio.run(main())

Configuration

The package relies on OpenAI or Azure OpenAI for LLM-based analysis. Set the following environment variables:

• OPENAI_API_KEY: Your OpenAI API key.
• Optional (for Azure): AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, AZURE_DEPLOYMENT_NAME.

Requirements

• Python 3.9+
• ffmpeg
• OpenCV
• faster-whisper
• LangChain