Home Movie AI

A Python script to automatically process home movies by:

• Detecting scene changes
• Extracting and transcribing audio using OpenAI's Whisper model
• Assigning transcript segments to corresponding scenes
• Generating a CSV file with scene numbers, timestamps, and transcripts

Features

• Scene detection using PySceneDetect
• Audio extraction with FFmpeg
• Speech-to-text transcription using Whisper

Requirements

• Python 3.8 or newer
• CUDA-enabled GPU recommended for performance (Whisper transcription)

Installation

pip install scenedetect openai-whisper torch torchvision torchaudio

Ensure you have FFmpeg installed and accessible from the command line.

Usage

Run the script with the video filename as a command-line argument:

python home_movie_ai.py "your-video-file.mp4"

Or simply:

python home_movie_ai.py

The script will prompt for the filename if not provided in the command.

Output

The script generates a CSV file named results-[your-video-file].csv with the following columns:

• Scene Number
• Start Time (min:sec)
• End Time (min:sec)
• Transcript

License

This project is licensed under the MIT License. Please provide attribution when using or modifying this software.

Attribution

Created by Ted Barnett. Contributions and forks are welcome!