AI disclosure

We're upfront about how the catalogue is made.

The music

Every track is generated by an instrumental music model from a written prompt and a fixed set of constraints (no vocals, no choirs, no field recordings). For each album, we author a per-track prompt that names the genre, mood, instrumentation, BPM range, and musical key, then run the prompt through the model and review the output. The result is what you hear on the site.

We don't sample, splice, or upload anyone else's recordings.

The cover art

Album and track cover art is generated by an image model with a per-genre style block that locks the aesthetic. We don't use face-bearing portraits and we deliberately keep the imagery abstract, illustrated, or environmental.

What gets stored

Each MP3 is tagged with standard ID3 frames (title, artist, album, BPM, key, genre, mood) and includes a "content provenance"marker noting it's AI-generated. The site itself is built and maintained by humans, and curatorial decisions (which albums to ship, which prompts to revise, which tracks to publish) are made by us, not by a model.

Why we tell you this

Music made this way is still relatively new and the conventions around disclosure are evolving. We'd rather over-disclose than have you feel surprised. If you have questions, write to us at hello@ravan.one.