Apple Music ai_transparency: Composition vs Track for AI songs
Apple Music spec 5.3.26 has Composition for AI lyrics or components and Track for an AI sound recording. A song from a prompt usually needs both values.

In the Apple Music specification 5.3.26, Composition means AI generated a material portion of the music composition (lyrics or other components), and Track means AI generated a material portion of a sound recording. A song generated end to end from a text prompt plausibly falls under both, and you decide that when you deliver.
Values are quoted from the Apple Music Specification page (April 2026), read 2026-10-01. Sume's side is the Music 1.0 and Music Router docs.
What do the Apple values mean?
The tag is <ai_transparencies> with one <ai_transparency> child per value, optional and updatable. If omitted, none is assumed. Track is available at track level only and applies to every territory; you disclose it for each track.
| Value | Meaning on the page |
|---|---|
Composition | AI generated a material portion of any music compositions in a track, such as lyrics or other components |
Track | AI generated a material portion of a sound recording; disclosed for each track |
Music Video | AI generated a material portion of the visual elements of a music video |
Artwork | Album-level value for AI-generated artwork |
Which values fit a song made from a prompt?
Sume's Music 1.0 docs say it runs on Google Lyria 3.5 and produces full-length structured songs steered by the prompt, with section markers such as [0:00-0:30] Intro: .... If the prompt also carried your own lyrics, the composition question gets more nuanced: the lyrics are yours, but the structure and arrangement came from the model. The spec text speaks of lyrics or other components, so judge each part separately.
The recording itself is generated audio, which points to Track. Check the spec's wording against your release rather than relying on this summary.
Can a Sume job hold my disclosure note?
Yes, as your own record. The metadata request field is described as "Caller metadata stored on the job; not sent to the provider." Put something like your release id and the values you plan to declare. It does not create the Apple tag; you still write that into your delivery metadata.
curl -X POST https://api.sume.com/v1/music-router/generate \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt":"a 2-minute synth-pop track","metadata":{"release":"EP-07","apple_ai_transparency":["Composition","Track"]}}'Which endpoint should new work call?
The Music 1.0 docs say it is retiring gradually, its routes keep working, and new integrations should call POST /v1/music-router/generate. The metadata field behaves as on Music 1.0 per the Router docs.
What should I do before delivery?
Record, per track, whether AI produced a material portion of the composition and of the recording, then add the matching values. Distributors may map this differently, so confirm with yours. See Lyria 3.5 lyrics and vocals for what the model generates. This is not legal advice.
Sources
Related posts
More in Use cases
- Amazon Responsive eCommerce Creative: assets to supply
Amazon's REC builds ads from lifestyle images, video, logos and copy. Prepare lifestyle stills with reference-guided edits and short clips.
- Android Automotive Play screenshots: 800x1280 and 1024x768
Play wants 2 portrait (800x1280) and 2 landscape (1024x768) Automotive screenshots of generic system UI. Both fit the ChatGPT Image 2.5 custom-size limits.
- Android XR Play screenshots: 8:5 at 3200x2000, not 3840x2400
Play recommends 3840x2400 for Android XR, but that is 9,216,000 pixels, over the ChatGPT Image 2.5 custom-size cap. Use 3200x2000: 8:5 and under it.
- Apple Podcasts video uses HLS: what Sume renders instead
Apple Podcasts video is delivered by HLS through hosting providers. Sume renders an MP4 episode with Timeline; it does not produce HLS.
Written by Sume