Turn long articles, videos, or course material into a private podcast to listen to on a commute.
Convert meeting notes into a spoken episode with your own custom host persona and voice.
Use the built in MCP endpoint to create podcast episodes directly from Cursor or Claude.
Review saved content later using the auto generated flashcards that come with each episode.
| vastsa/bokebox | codedgar/three-fenestra | futureuniant/workshadow | |
|---|---|---|---|
| Stars | 42 | 42 | 42 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | easy | moderate |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | general | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Requires your own AI model API keys and a self-hosted deployment via the start script or Docker.
BokeBox is a self-hosted tool that turns things you would normally read or watch, such as videos, web links, articles, meeting notes, or course materials, into private audio podcasts you can listen to instead. The idea is that you have saved a lot of content you keep meaning to get to, but you never find the quiet, uninterrupted time to sit and read or watch it. BokeBox lets you listen to it during a commute, chores, or a walk instead. To make an episode, you drop in a source such as a video file, a link, an article, or a draft, and the system can be extended with plugins to accept more source types. It reads and understands the content, rewrites it into a script meant to be spoken rather than read aloud word for word, and then generates audio in a voice you choose. You can also define a host persona: who the host is, their tone, how they open and close a show, and the show's name, either as a global default or set differently per episode. Each episode also comes with automatically generated flashcards so the main points are easy to review later. The project includes a built in MCP endpoint, so tools like Cursor or Claude can create podcast episodes or check on job status directly, using tools such as create_podcast_from_url and create_podcast_from_text. This makes it possible to trigger podcast creation from an AI coding assistant rather than only through BokeBox's own interface. BokeBox is designed for a single user running their own private deployment, so your source content, tasks, and listening progress stay on your own machine rather than living on a shared platform or public feed. The README is clear that it is not meant to be a public podcast platform, a multi-tenant service, or a tool for live collaborative editing. To get started, you clone the repository, copy an example environment file and add your own API keys, then run a start script, or alternatively pull a prebuilt Docker image. The first launch walks you through account setup and configuring which AI model to use. The project is licensed under LGPL-3.0.
A self-hosted tool that converts videos, links, articles, and notes into private AI-narrated podcast episodes with a custom host persona and voice.
Mainly TypeScript. The stack also includes TypeScript, MCP, Docker.
You can use, modify, and share the software, including in your own projects, but if you modify the library itself and distribute it, you must share those changes under the same license.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
double-check against the repo, no cap.