PodGrab

A command-line tool for downloading podcast episodes and generating transcriptions using OpenAI's Whisper model.

Features

• 🔍 Search podcasts using the iTunes Search API
• ⬇️ Download individual podcast episodes
• 🎯 Interactive episode selection for ambiguous matches
• 📝 Transcribe audio files using OpenAI's Whisper model
• 📊 Progress bar for downloads
• 📋 Support for different Whisper model sizes (tiny to large)
• 🔄 Non-interactive mode for automation

Installation

From PyPI (Recommended)

pip install podgrab

From Source (For Development)

1. Make sure you have Python 3.11 or later installed
2. Install Poetry if you haven't already:

   pip install poetry
   

3. Clone this repository and install dependencies:

   git clone https://github.com/username/podgrab.git
   cd podgrab
   poetry install
   

Usage

Downloading Podcasts

Basic usage to download a podcast episode:

podgrab download "Podcast Name" "Episode Title"

Options:

• --output/-o: Specify output directory (default: current directory)
• --non-interactive: Run without interactive prompts
• --verbose/-v: Enable verbose output
• --quiet/-q: Suppress non-error messages

Example:

podgrab download "The Daily" "Today's Episode" -o ~/Downloads

Transcribing Audio

Transcribe a downloaded podcast episode:

podgrab transcribe path/to/audio/file.mp3

Options:

• --model/-m: Specify Whisper model size (tiny, base, small, medium, large)
• --output/-o: Save transcription to file (default: print to console)

Example:

podgrab transcribe podcast.mp3 -m medium -o transcript.txt

Requirements

• Python 3.11+
• Required Python packages (automatically installed):
  • typer
  • requests
  • feedparser
  • rich
  • tqdm
  • openai-whisper
  • torch
  • torchvision
  • torchaudio

Development

1. Set up the development environment:

   poetry install
   

2. Run tests:

   poetry run pytest
   

Contributing

1. Fork the repository
2. Create a new branch for your feature
3. Make your changes
4. Add or update tests as needed
5. Submit a pull request

License

This project is licensed under the MIT License.

Acknowledgments

• Uses the OpenAI Whisper model for transcription
• Built with Typer for the CLI interface
• Uses the iTunes Search API for podcast discovery