About this project

Podcastfy is an open-source Python package that converts multimodal content into conversational audio podcasts using generative AI. It is positioned as an open, programmatic alternative to closed UI-based tools for research synthesis, emphasizing customization and scale rather than a single hosted interface. Inputs and outputs - Accepts websites, PDFs, images, YouTube videos, and user-provided topics as source material. - Produces multi-lingual audio conversations, with support for short (roughly 2-5 minute) and longform (30+ minute) formats. - The README shows example outputs generated from images, personal websites, long interviews, books, and non-English news or research pages. How it is used - Installable from PyPI as `podcastfy`; requires Python 3.11+ and ffmpeg for audio processing. - Python usage is a single call: `generate_podcast(urls=[...])` from `podcastfy.client`. - A CLI is available via `python -m podcastfy.client --url ...`. - A beta FastAPI/Docker path is documented for serving URL-based requests. Customization and integrations - Conversation format, style, and voices can be customized. - Transcript generation can use 100+ LLM models from providers such as OpenAI, Anthropic, and Google, or local LLMs (the README mentions 156+ HuggingFace models) for greater privacy and control. - Text-to-speech integrations include OpenAI, Google, ElevenLabs, and Microsoft Edge. - Configuration and API keys are handled through documented settings files. Project status and ecosystem - Licensed under Apache 2.0, with documentation on Read the Docs, a Colab notebook, tests, and Docker publishing workflows. - The README lists projects built with Podcastfy, including OpenNotebook, SurfSense, OpenPod, Podcast-llm, and a Hugging Face demo. - Stated use cases include content creators repurposing articles into audio, educators converting lecture materials, researchers producing audio summaries, and accessibility advocates turning written or visual content into auditory formats. This overview reflects only capabilities described in the repository README; no independent benchmarks, rankings, or performance claims are made here.