Skip to content

🎙️ P³ v0.1.0: Initial Release

Latest

Choose a tag to compare

@haasonsaas haasonsaas released this 25 Aug 18:10
· 6 commits to main since this release

Parakeet Podcast Processor (P³) v0.1.0

The world's fastest podcast processing system for Apple Silicon

🚀 What's New

This is the initial release of P³, a revolutionary podcast processing system that leverages:

  • ⚡ Parakeet MLX: 30x faster transcription than Whisper on Apple Silicon
  • 🧠 Ollama Integration: Local LLM analysis with structured data extraction
  • 📥 Smart Audio Pipeline: RSS monitoring + ffmpeg normalization
  • 💾 Efficient Storage: DuckDB for fast analytical queries
  • 📊 Rich Outputs: Markdown and JSON digest generation
  • 🔧 Professional CLI: Complete command-line interface

📈 Performance

  • Complete Pipeline: ~2 minutes for full podcast processing
  • Audio Download: ~40 seconds per episode
  • Parakeet Transcription: ~28 seconds (lightning fast!)
  • Ollama Analysis: ~46 seconds for structured summarization
  • Export: <1 second

🛠️ Quick Start

# Prerequisites: macOS + Apple Silicon + ffmpeg + Ollama
brew install ffmpeg
# Install Ollama from https://ollama.com, then: ollama pull llama3.2

# Setup P³
python3 -m venv venv && source venv/bin/activate
pip install -e .
p3 init

# Configure feeds in config/feeds.yaml, then run:
p3 fetch && p3 transcribe && p3 digest && p3 export

🏗️ Architecture

RSS → ffmpeg → Parakeet MLX → Ollama → DuckDB → Export

100% Local Processing - No API keys required, complete privacy.

🔗 Links

Perfect for researchers, journalists, content creators, or anyone processing large volumes of podcast content.