AI may not be our friend when it comes to music creation. But I spent this afternoon simply chatting with Claude ai, testing and revising a Roon remote extension it created to help block AI content. Nowhere near foolproof, but a good start if you ask someone like me, who has no computer coding experience.
Here is an overview of what Claude ai created and some screenshots:
AI Music Detector — Roon Extension v1.6.1
For those who prefer to keep AI-generated music out of their Roon listening experience — particularly from Roon Radio pulling tracks from TIDAL and Qobuz — I wanted to share an extension I’ve been developing with some help.
What it does:
The AI Music Detector is a Node.js Roon extension that monitors all playback in real time and automatically skips or flags tracks identified as AI-generated. It works regardless of how a track was queued — whether you played it manually or Roon Radio selected it. When a track is detected, a Windows desktop toast notification pops up identifying what was blocked.
Three-layer detection:
- Built-in blocklist — 30 known AI music platforms and services that distribute under their own name on streaming services (Endel, Mubert, Suno, Udio, ElevenLabs, Boomy, AIVA, Soundraw, and more)
- Soul Over AI community database — automatically fetches the community-curated list of ~1,375 verified AI artists from the Soul Over AI GitHub repository on startup, refreshing daily. This catches disguised AI artists distributed under individual names that the built-in list would miss entirely
- Custom blocklist — a settings field in Roon where you can add any AI artist names you discover personally, taking effect immediately without restarting anything
Settings available in Roon → Settings → Extensions:
- Pause the detector temporarily to intentionally listen to an AI track
- Switch between auto-skip and flag-only modes
- Add custom AI artist names, label keywords, and genre keywords
Setup requirements:
- Windows PC on the same local network as your Roon Core
- Node.js (free, from nodejs.org)
- Works alongside Nucleus+, ROCK, or any other Roon Core — the extension runs on your Windows PC and communicates with the Core over your local network
- Starts automatically at Windows login via Task Scheduler — no Command Prompt window needed during normal use
Known limitations — please read before using:
- Detection is metadata-based only. The extension checks artist names, album names, and genre keywords against its blocklists. It does not perform any audio analysis.
- Disguised AI artists are the hardest problem. When Suno or Udio creators distribute music to TIDAL/Qobuz under a personal artist name with no AI disclosure in the metadata, detection depends entirely on whether that artist appears in the Soul Over AI database or your custom list. If an AI artist isn’t in either list, the extension won’t catch them.
- Soul Over AI database is frozen. The Soul Over AI project has sunset and is no longer actively maintained, meaning the community database won’t grow. It remains a valuable frozen snapshot of ~1,375 verified AI artists, but new AI artists emerging after the shutdown won’t be added automatically.
- Genre keywords may cause false positives. Broad keywords like “royalty free” or “production music” could occasionally flag legitimate human-made music. If you experience false positives, the custom blocklist fields can be used in reverse — let me know and I can add a whitelist feature in a future update.
- Windows only for toast notifications. The core extension works on any platform running Node.js, but the desktop notification feature is Windows-specific. Mac/Linux users can still use the extension — detection and auto-skip work fine, just without the popup notification.
- Qobuz and TIDAL’s own AI labeling is incomplete. Even though Qobuz has an AI Charter and detection tool, AI music is still widely present in both catalogs and surfaces freely through Roon Radio. This extension fills that gap for Roon users specifically.
What it cannot do:
- Detect AI music by audio fingerprinting or acoustic analysis — no reliable public API exists for this yet
- Catch AI artists whose streaming metadata contains no identifying information and who aren’t in any known database
- Work on Roon Nucleus/ROCK directly — the extension must run on a separate machine on the same network
I’d welcome feedback, additional AI artist names to add to the built-in list, and any interest from developers in building a maintained successor to the Soul Over AI database specifically for streaming platform AI artists. That would make this extension significantly more powerful over time.
