RETROSPECTIVE RECORD · PREPARED 16 SEPTEMBER 2026The archive · 100 retrospective records ↗
← The archive

Two voice-conversion services now publish artist consent as a feature

Kits AI and Voice-Swap market artist-approved voice libraries, but their own terms put the consent burden for uploaded voices on the user.

Site publication
September 18, 2026
Visual for this record: Two voice-conversion services now publish artist consent as a feature
Visual published by framerusercontent.com, shown for identification of the record. Credit: framerusercontent.com · source page ↗ Rights: owner-review-pending. Source

What happened

Kits AI and Voice-Swap are two commercial voice-conversion services that, as retrieved on 16 September 2026, market themselves around artist consent rather than open cloning. Kits AI's own homepage states that 'every voice used in our models is ethically licensed and securely sourced by Kits via the artists themselves', and describes a programme called Kits Earn through which artists who contribute a voice model receive a share of its use. Voice-Swap's business page describes its custom voice offering as built 'from authorised recordings' with rights 'secured' and models 'contract-backed from inception'. Both frame licensed sourcing as their point of difference from open-source cloning tools.

What the documents say

Kits AI's terms of service distinguish a pre-built library of licensed voices from custom models a user creates from their own uploaded audio. For custom models, the terms require the user to represent they hold 'all rights, licenses, consents, permissions' needed for any voice file they submit, and separately prohibit using the service to 'impersonate somebody... without their authorization'. The terms do not themselves document how the pre-built library's artist licences were obtained; that claim appears only on the marketing homepage, not in the contractual document. Voice-Swap's business page similarly asserts that ownership and licensing are 'confirmed in writing before you commit', without publishing the licence text or a royalty figure.

Why it matters for makers

The mechanism worth separating here is marketing claim versus contractual obligation. A platform's homepage can describe an ethics programme while its terms of service place the actual, enforceable consent burden mainly on the end user for anything they upload themselves — the pre-built library's clearance is asserted, not demonstrated, in the document a user actually agrees to. That gap matters commercially: if a rights dispute arose over a library voice, the user's protection depends on representations made outside the contract they signed.

What to check before you use it

Before releasing a commercial track built on a pre-built library voice, a maker should ask the platform directly for the licence or consent record behind that specific voice, rather than relying on homepage language; confirm what 'ethically licensed' covers — the recording, the likeness, or both; and, for any custom model built from someone else's voice file, keep the uploader's own written consent, since the terms of service assign that responsibility to the uploader, not the platform.

Ethically-sourced marketing language is a reasonable signal of intent, but the enforceable document is the terms of service, and on both services reviewed here that document is narrower than the homepage claim sitting beside it.

Sources & reading trail

The contractual terms: custom-model ownership assignment, output licence, and the user's consent representations.

Source published: Not established · Retrieved: 16 September 2026

The company's own marketing claim that library voices are 'ethically licensed' and its Kits Earn revenue-share programme.

Source published: Not established · Retrieved: 16 September 2026

The platform's own description of licensing custom voices from authorised recordings, contract-backed from inception.

Source published: Not established · Retrieved: 16 September 2026

Papers, terms and official documents establish the record; the maker reading and the checks are Signal to Song editorial analysis. This retrospective draft does not imply the site published on the event date.

Continue reading

Sources & reading trail

The documents above establish the record. The reading and the questions are this publication’s editorial analysis, written after the fact.

Published September 18, 2026, not on the date of the event described.