Composition
You wrote it, or co-writers with an interest in it agree to this use.
Trust
Your music stays yours
ANYANO only works if artists are willing to upload material that matters to them. This page states plainly what rights you need, what happens to your uploads, and where the limits of our promises are.
Requirements
Before you upload: the rights you need
Training on your own catalog is the whole premise. That only holds if you actually control the material.
Recording
You own the master, or the rights holder — often a label — permits this use.
Performances
Session players, featured artists and vocalists on the recording permit it.
Samples and clearances
Any third-party sample, loop or interpolation in the material is cleared for this use.
ANYANO asks you to confirm you hold these rights before a training run starts, with a separate confirmation for voice recordings. That confirmation is yours to make honestly — we cannot verify your agreements for you.
Handling
What happens to uploaded music
- 01
Stored privately
Files go into a private storage bucket, not to public URLs. Access is authenticated and scoped to your account; there are no guessable public links to your audio.
- 02
Isolated per account
Database rows for files, datasets, Sounds, jobs and generations are protected by row-level security policies tied to your user id, so other accounts cannot read them.
- 03
Processed to run your jobs
Analysis, stem separation, preparation, training and generation require processing your material — including on connected specialist compute configured for your studio.
- 04
Not pooled into shared models
Your material trains your Sounds. It is not used to train Sounds for other customers.
Rights in results
Generated output
Output from your Sound is generated from material you provided, and the Terms of Service define the rights you hold in it. Two honest caveats belong here rather than in small print:
First, copyright protection for material generated without sufficient human authorship is unsettled and varies by jurisdiction — several copyright offices have declined to register purely machine-generated work. Second, if your training material was not fully cleared, problems in the input can follow the output. Neither risk is removed by using ANYANO.
The practical approach: treat generations as raw material, do real creative work on top of them, and take advice from qualified legal counsel before a commercial release that leans on generated content.
Control
Deletion
You can delete uploaded files from your library, and delete Sounds and versions you have trained. Deleting a file does not retroactively untrain a Sound that already learned from it — if you want that material out of your model, delete the affected Sound version as well and retrain from a dataset that excludes it.
Retention windows, backup handling and processor-side deletion are covered on the privacy page for artists.
Honesty
Where our claims stop
We will not tell you your music is "100% private" as a slogan. What we will tell you is specifically what is implemented: private storage, per-account access policies, no cross-account training, rights confirmation before training, and deletion controls in the app. Anything that depends on a third-party processor is described as such on the privacy page.
Where this page and the Terms of Service differ, the Terms govern. This page is an explanation, not legal advice.
FAQ
Frequently asked questions
Related
Keep reading
Train on your own terms.
Confirm your rights, upload your music, keep control of your Sound.