YouTube Audio Transcription Script

Overview

yt_scribe.py is a command-line tool that downloads audio from YouTube videos, transcribes the audio using OpenAI's Whisper model, and exports both the transcription and relevant metadata as files.

Features

• Audio Download: Extracts audio from YouTube videos.
• Automatic Transcription: Transcribes audio using Whisper with GPU/CPU support.
• Language Auto-Detection: Detects language automatically if not specified.
• Metadata Export: Saves video metadata (title, channel, publish date, and detected language) to a JSON file.
• Customizable Output: Configurable model size, language, and output directory.

Requirements

• Python 3.8+
• Libraries:
  • torch
  • whisper
  • yt-dlp
  • argparse
  • json

Installation

1. Download the script.
2. Install dependencies:

   pip install torch whisper yt-dlp

Usage

Basic Command:

python yt_scribe.py -u "<YouTube_URL>" -o <output_directory>

Command-Line Arguments:

Argument	Description	Default
-u, --urls	Comma-separated YouTube URLs or file path	Required
-o, --output_dir	Output directory	Current directory
-m, --model_size	Whisper model size (tiny, base, small, medium, large)	base
-l, --language	Language code (e.g., en, es) or auto-detection	Auto-detect

Example:

python yt_scribe.py -u "https://www.youtube.com/watch?v=dQw4w9WgXcQ" -o transcriptions/ -m base -l en

Processing a List of URLs:

python yt_scribe.py -u youtube_urls.txt -o transcriptions/

Output Files

1. Transcription File: <video_title>_transcription.txt
2. Metadata File: <video_title>_metadata.json

License

This project is licensed under the MIT License. See the LICENSE file for details.

Contact

For any questions, reach out at my mail.