Blog

The Most Unexpected CKAN Use Case I've Ever Seen: A Sheet Music Directory With AI Metadata

Wolfgang from Ondics built an open source sheet music catalog on CKAN β€” with AI metadata generation, YouTube playback, and cross-instance sharing. Here's how.

Someone Built a Sheet Music Directory on CKAN. I Did Not See That Coming.

I've been hosting CKAN Monthly Live for a while now. Government portals, humanitarian data platforms, open data infrastructure at national scale. I thought I had a solid sense of what people build with CKAN.

Then Wolfgang Clauss from Ondics joined the call from Germany and said he had something to show us.

He was right that none of us had thought about it. Wolfgang and his team built a fully functional open source sheet music directory on top of CKAN β€” with AI-assisted metadata generation, inline YouTube video playback, licensing protection for PDF scores, and cross-instance collection sharing. No existing open source metadata management solution for musicians existed. So they built one. And they're giving it away.

Watch the demo Wolfgang Clauss, Ondics ← From CKAN Monthly Live #41 β€” full recap
01

The problem: a musician drowning in metadata

Wolfgang has a colleague who is, in his words, "a heavy musician" β€” someone who plays many instruments and is constantly managing sheet music. The challenge isn't finding music. It's organising it. Knowing what you have, what instrument it's written for, who composed it, which part you're playing, whether you're allowed to share it. Sheet music comes with a lot of metadata, and there was no good open source tool to manage it at scale.

Wolfgang and his team at Ondics knew CKAN well. They looked at the problem and saw an obvious fit: CKAN's dataset and resource pattern maps directly onto music. A dataset is a song. A resource inside that dataset is a PDF score. The metadata schema is extensible. The API is already there. The harvesting, the search, the user management β€” all of it.

"Metadata management is more important than ever β€” and we have not even discovered how important metadata management is right now."

β€” Wolfgang, Ondics, CKAN Monthly Live #41

The result is a CKAN extension that turns any CKAN instance into a searchable, filterable sheet music library. Browse by song title, instrument, composer, or style. Open a piece and the sheet music loads alongside a matching YouTube video so you can hear what you're about to play. It is, as Wolfgang put it, "optimised for doing everything a musician likes."

Sheet music homepage – CKAN compact
02

The features that made the chat go quiet

Wolfgang walked through the extension live. A few things stood out.

Licensing and access control

Sheet music is protected by copyright. The extension enforces this properly β€” PDF scores are locked behind login. Unauthenticated users see the metadata but cannot open the file. It's a clean solution to a problem that most music software handles badly or not at all.

Extended musical metadata schema

The standard CKAN schema is extended with instrument, part, composer, and style. When you download a PDF, the music metadata is embedded in the file itself β€” so it can be imported directly into mobile sheet music apps that display scores while you're performing. The data travels with the music.

Sheet music PDF viewer – CKAN compact
Export to Dropbox for mobile performance

Search for a set of pieces, export the results directly to Dropbox. Your mobile music app picks them up from there and handles display during performance. Wolfgang showed this as if it were obvious β€” it isn't. It's a complete end-to-end workflow that would take months to build in a dedicated music application.

Live YouTube video research

From inside a dataset, you can run a live YouTube search based on the song. Results return in real time. You select the video that best matches your version or your preferences, and it's saved to the dataset permanently. Your sheet music and its reference recording are linked.

Research video – CKAN compact
AI metadata generation

This is the feature that got the biggest reaction. When uploading a new piece, you enter the title. One click sends it to Google Gemini or OpenAI ChatGPT with a structured prompt requesting composer, style, and a data description. The AI returns suggestions. One more click saves them directly to the CKAN dataset.

What used to require five minutes of research per piece β€” composer dates, style classification, descriptive text β€” becomes a ten-second operation. For a musician managing hundreds of pieces, that compounds fast.

"You don't have to hassle about typing in information about this piece of music. You enter a title, you execute the prompt, and the metadata is added to the dataset."

β€” Wolfgang, Ondics, CKAN Monthly Live #41

Research metadata – CKAN compact
03

Collections, songbooks, and sharing between musicians

The extension reimagines one of CKAN's core concepts. In standard CKAN, organisations are used to group datasets by publisher or owner. In this extension, organisations become collections β€” essentially digital songbooks. Wolfgang demonstrated one called "Piano Piano" containing 100 pieces.

Each collection has an export feature that packages every PDF it contains into a single tar or zip file. That archive can then be imported directly into any other CKAN instance running the same extension. A music school in Berlin can share a curated repertoire with an ensemble in Vienna. An orchestra can distribute parts to individual musicians. No manual re-entry. No re-uploading. No emailing PDFs one by one.

Then came the moment I want to highlight β€” not a feature, but a statement. After showing everything, Wolfgang said this:

"I didn't see any metadata management for musicians on the market. This is no business case for us. We will open this to the public in short. It's simply one extension with some configuration and that's it."

β€” Wolfgang, Ondics, CKAN Monthly Live #41

They identified a real gap. They built a solution on CKAN because CKAN gave them the solid foundation they needed. And they're handing it to the community. That's the open source spirit that makes this ecosystem worth being part of.

Get involved

Wolfgang ended his demo with: "We had a lot of fun doing this extension. In CKAN we saw a very solid basis for this kind of metadata management β€” AI integration, export, the assistance of musicians during playing."

What this use case proves is that CKAN's architecture is far more flexible than most people assume. The same platform that powers national government data portals can run a sheet music library with AI metadata and cross-instance sharing.

If you're building something unexpected on CKAN β€” however niche, however far from "open government data" β€” we want to hear about it. The CKAN community forum is the place to share what you're working on. And if you'd like to present at a future CKAN Monthly Live, fill in the speaker form β€” the more unexpected the use case, the better.

Sheet music search results – CKAN compact
Hello Dolly dataset – CKAN compact

Further reading
There Was a Lot More at CKAN Monthly Live #41 β€” Read the Full Recap
The complete recap: CKAN Ecosystem Catalog, HDX Tabular Data Endpoints, new community forum, and this sheet music use case.
β†’
Sheet music directory on CKAN β€” Wolfgang Clauss, Ondics, CKAN Monthly Live #41
Watch the sheet music demo β€” Wolfgang Clauss, Ondics
Watch on YouTube β†’