AI transparency

All music in Meus is generated by AI, on your iPhone. Here is what that means.

Which model makes the music

Every track is composed by Stable Audio 3 Small, an open music model by Stability AI, used under the Stability AI Community License. The model is built into the app and runs on the device. The first tracks of each station come with the app, made in advance by the same model so the music starts at once; everything after them is composed on your iPhone. No track is streamed or written by a person. The ambience sounds (rain, fire, waves, forest) were also generated with Stable Audio.

What the AI does

• Each station is a short text description of a genre, mood, tempo and instruments. The model turns it into new instrumental music, piece after piece.
• Your own station works the same way, from the words you type. They stay on your iPhone.
• The model does not learn from you and nothing you do is sent anywhere.

How it is labeled

• In the app, the music is shown as AI-generated.
• Files you export from Meus will carry metadata saying they were generated by AI, in a machine-readable form. TODO: export metadata is planned, not built yet
• This page, the App Store description and the site footer say the same.

No imitation of real artists

The stations are described by style, never by the name of a real artist, and we don't market any station as sounding like a specific person. The Terms ask you not to use your own stations to imitate real artists.

EU AI Act, Article 50

Article 50 of the EU AI Act requires that synthetic audio made by AI systems is marked as artificially generated in a machine-readable way, and that people are told when content is AI-generated. The in-app labels, the file metadata and this page are how Meus does that.

Questions

Write to support@meus.fm.