Skip to content

README.md

Latest commit

 

History

History
56 lines (37 loc) · 1.1 KB

README.md

File metadata and controls

56 lines (37 loc) · 1.1 KB
Raw

Podcast AI Agent

Download and transcribe audio from YouTube videos using OpenAI's Whisper model.

Prerequisites

  • Python 3.10+
  • uv (Dependency manager)
  • ffmpeg (Required for audio extraction)

Installation

  1. Clone the repository.
  2. Install dependencies:
make install

Usage

To run the agent:

make run

Or execute the CLI entry point directly via uv:

uv run podcast-ai-agent

Configuration

Configuration is managed via config/default.yaml and environment variables. Key settings include:

  • Whisper: Model size (tiny, base, small, medium, large-v3), language.
  • Download: Audio format, codec, timeout.
  • Output: Directory, file format (txt, json, srt, vtt).

Development

We use uv for package management and ruff/black for linting/formatting.

Run all checks:

make check

Common commands:

  • make format: Auto-format code.
  • make lint: Check for linting errors.
  • make test: Run unit tests.
  • make type-check: Run static type analysis.