Odyon — AI Music Studio

Synthetic voices should amplify creativity — never impersonate, deceive, or exploit. These are the rules that keep Odyon's vocal tools on the right side of that line.

1. Purpose

Odyon's vocal features — Voice Swap (converting a vocal performance into another voice) and Vocal Studio (rendering sung vocals from melody and lyrics) — are powerful creative tools, and with that power comes responsibility. This policy sets out the ethical rules that govern how these features may be used, and the commitments we make about the voices we offer.

2. Curated Voices Only

Every voice available in Odyon's vocal features comes from a curated catalog that we control. No voice in the catalog is offered as, or presented as, any specific living person. We are completing a licence and provenance review of every model in the catalog, and we will withdraw any voice whose chain of rights we cannot stand behind. Odyon does not allow users to add, upload, or train custom target voices. You cannot clone a voice on this platform — the conversion targets are limited to the catalog we provide. This is a deliberate design decision, not a temporary limitation.

3. Prohibited Uses

You may not use Odyon's vocal features to: - Upload recordings of another person's voice — including recognizable artists — without their written authorization, for any purpose. - Impersonate any real person, or present AI-generated vocals as an authentic recording of a real person. - Create content for fraud, scams, harassment, defamation, or deception of any kind. - Produce deepfake-style audio of public figures, including political disinformation or fabricated statements. - Circumvent, or attempt to circumvent, the curated-catalog restriction. Violation of this section results in immediate account termination without refund of subscription fees or remaining Gems, and — where required — cooperation with law enforcement.

4. Your Uploads & Warranties

When you upload a vocal for Voice Swap, you represent and warrant that you recorded it yourself or otherwise hold all necessary rights to it, and that processing it does not violate any third party's rights. Source vocals you upload are retained alongside the converted output for provenance — so every AI-converted vocal on the platform can be traced back to its original input. This retention supports abuse investigation and rights-holder complaints, and is described in our Privacy Policy.

5. Provenance & Transparency

AI-converted and AI-synthesized vocals are labeled as such within the platform (for example, by their type in your library and in the Explore gallery). In addition, audio our servers render for you — generated music and sound effects, voice conversions, and Vocal Studio renders and mixes — carries a machine-readable marker in its embedded metadata identifying it as AI-generated (an INFO block in WAV files, an ID3 tag in MP3 files), naming the generator and, for vocals, the voice used and its provenance. Some files are assembled in your browser rather than on our servers — chop kits sliced from a vocal, and project mixdowns exported from the studio — and those do not carry the marker yet; we are extending it to them. This is how we meet the transparency obligation for synthetic audio under Article 50 of the EU AI Act. The marker is not applied to your own recordings — for example the source vocal you upload for a voice swap. We are also evaluating inaudible watermarking technologies for AI vocal outputs as the industry standardizes around them. When you release a vocal produced by our spoken-vocal engine, you must make clear to your listeners that the voice is AI-generated and not a human voice. That is not our preference — it is a condition of the licence under which we provide that engine, and it passes to you. For every other AI-generated vocal we ask you to follow the disclosure norms of the platforms you distribute on, which increasingly require the same thing.

6. Enforcement & Reporting