RSS Feeds for (almost) every ARDSounds Podcast

Why?

I like listening to podcasts from the German public broadcasters, but I like listening to them in my podcast app. ARDSounds (formerly the ARD Audiothek) has a huge catalogue, yet many shows don’t come with a public RSS feed you could simply paste into your app of choice. When I checked, only about a quarter of the shows had one from their broadcaster.

So I wrote a little tool that builds those feeds myself. It started as a Python reimplementation of the PHP project ARD-Audiothek-RSS by matztam and slowly grew from “a feed for my favourite shows” into “a feed for every show”.

👉 The result lives here: gboeer.github.io/ardsounds-feeds

Using the feeds

No account, no app, no setup:

1. Open the feed page and search for a show (or a station like “Bayern 2” or “WDR 5”).
2. Hit the copy button next to the feed URL.
3. Paste it into your podcast app’s “add podcast by URL” dialog. Done.

A few extras:

• Lots of shows at once: every station has its own OPML file, and there’s one with all podcasts. Most podcast apps can import OPML, so you can subscribe to a whole station in one go.
• Official feeds: if the broadcaster does offer its own feed, it’s linked as “Offizieller Feed” next to the show, in case you prefer the original.
• Fresh every morning: all feeds are rebuilt daily and contain the 100 newest episodes of each show.

How it works (the short version)

There’s no server involved. A GitHub Actions workflow runs once a day and:

• asks the ARD’s GraphQL API for every show with at least one episode,
• builds an RSS feed for each of them (episodes whose audio file has vanished are dropped, since your app couldn’t play them anyway),
• renders the overview page and the OPML files,
• and publishes everything as static files on GitHub Pages.

Feeds from the previous run are cached, so a normal daily run only has to look at what’s new and takes a few minutes. If ARD has a hiccup with one show, the last working version of its feed simply stays online. Everything is plain Python standard library, no dependencies.

Make it your own

The code is open source (GPL-3.0) at github.com/gboeer/ardsounds-feeds, and you can run your own copy without ever leaving the browser.

Fork it on GitHub

1. Fork the repository.
2. In your fork, go to Settings → Pages and set the source to GitHub Actions.
3. Open the Actions tab. GitHub disables workflows in fresh forks, so enable them there.
4. Start the workflow “Fetch Podcast RSS Feeds” with Run workflow. The very first run takes around a quarter of an hour, because it has to build every feed from scratch.

Your feeds then show up at https://<your-username>.github.io/ardsounds-feeds/, and they keep updating every day on their own.

Tweak the settings

All settings live in podcasts.json, which you can edit right on GitHub (pencil icon, commit, done):

{
  "default_latest": 100,
  "exclude": ["<show_id>"],
  "shows": [
    { "show_id": "8e6d4d6fa453e7f7", "name": "Kalk & Welk", "latest": 20 }
  ]
}

• default_latest: how many of the newest episodes go into each feed.
• exclude: shows you don’t want a feed for.
• shows: your favourites. They appear in their own “Ausgewählt” section at the top of the page and can get their own name or episode count. "latest": null means all episodes.

The show_id is the last part of a show’s URL on ARDSounds: for https://www.ardsounds.de/sendung/kalk-und-welk/urn:ard:show:8e6d4d6fa453e7f7/ it’s 8e6d4d6fa453e7f7. A small check runs on every change to podcasts.json, so a typo or a stray comma gets caught before the daily run.

Only interested in your own handful of shows? In .github/workflows/fetch-feeds.yml, change python fetch_feeds.py to python fetch_feeds.py --listed-only, and the workflow only builds the feeds listed under shows. The page itself (texts, colours, layout) lives in templates/.

Or run it locally

If you’d rather keep things on your own machine, clone the repo. All you need is Python 3:

git clone https://github.com/gboeer/ardsounds-feeds.git
cd ardsounds-feeds

# A single feed, printed to stdout (here: the 10 newest episodes of "Kalk & Welk")
python3 ardaudiothek_rss.py --show 8e6d4d6fa453e7f7 --latest 10

# A tiny local feed server: http://localhost:8000/?show=8e6d4d6fa453e7f7&latest=10
python3 ardaudiothek_rss.py --serve --port 8000

# The batch run for the shows in podcasts.json, plus the overview page
python3 fetch_feeds.py --listed-only
python3 build_pages_index.py

The local server is handy if you have a Raspberry Pi or a NAS at home: point your podcast app at it and you get feeds for any show, built on demand.

Wrapping up

It’s a small project, but it scratches an itch: all of ARDSounds in whatever podcast app I like. If you find a show that doesn’t work or have an idea, feel free to open an issue on GitHub. Happy listening! 🎧