Kits AI lets musicians clone their own voice, convert vocals with a library of royalty-free AI singers, and split, clean or master audio on the web or in a desktop beta.
Used this tool? Rate it
Used this tool? Rate it
Kits AI (styled Kits.AI on parts of its own site) is an AI vocal and audio toolkit for music production. It presents itself as a set of studio-quality AI audio tools for creating custom voices, singing in any style, playing any instrument and isolating vocals, all described as 100% royalty free.
The kits.ai site footer credits Arpeggi Labs Inc. as the copyright holder. The Terms of Service, last revised on October 2, 2024, describe the website as operated on behalf of Arpeggi, Inc. Trade press reported on January 21, 2026 that Splice acquired Kits AI and that co-founders Evan Dhillon and Kyle Dhillon joined Splice along with the Kits AI team; the site footer and the Terms still carry the Arpeggi names.
Kits AI centres on vocals: its generative feature produces short sung lines from text, while most of the platform works on audio you already have.
Kits Studio hosts the AI voice changer, which transforms any audio into a different voice while keeping the delivery of the original.
AI voice cloning on Kits comes in two routes that trade speed for accuracy.
Pages differ on how much audio a professional clone needs. The help article recommends 10–60 minutes of dry, monophonic vocals for Professional Voice Cloning. The voice cloning tool page instead says to prepare at least 10 minutes of dry, monophonic vocals before pressing Train.
The AI voice changer workflow in Kits Studio follows four steps.
Kits' own use-case pages centre on demos and pre-production vocals.
Kits positions its tools for vocalists, producers and content creators, and the fit differs by group.
It is a weaker fit for anyone under 18, who cannot use the service, for buyers who need a straightforward refund path, and for people who want to clone a voice they have no authority to use.
Text-to-speech was removed from the Kits AI API as of September 22, 2025. The homepage tool list still shows a Text-to-speech entry for generating audio from text with any voice. None of the official pages checked describes an iOS or Android app.
Kits uses a free tier plus three monthly subscriptions, and the plan cards, the pricing FAQ and other site pages do not fully agree.
Paid subscriptions include unlimited conversion time but charge download minutes when audio leaves the platform; those minutes refresh monthly, unused ones roll over, and unlimited minutes are subject to fair use.
The pricing page has an annual billing toggle labelled up to 47% off, but annual per-plan prices were not displayed in the captured default view. The starting price also differs across pages, as the homepage says paid plans start at $10 per month. The voice cloning tool page instead says paid plans start at $14.99 per month. Listed pricing applies to new users, while past users keep the pricing model that was active when they subscribed. Payments are supported only via card and other Stripe connections such as CashApp.
Kits' own focus is singing-voice conversion, a royalty-free AI singing voice generator library, and music-specific utilities such as stem splitting and mastering.
The biggest constraints sit in billing terms, clone quality and the rules around whose voice you may use.
In AI voice cloning, speed has a cost: Instant Cloning is not ideal for high-fidelity work or conversions that need high similarity to the target voice.
As captured on October 5, 2026, a third-party review platform showed a 1.6 rating from 55 company-level reviews of kits.ai, 91% of them one-star, and noted that the company had not invited reviews recently, so reviews may not be representative.
Outside Kits Earn, none of the help-center or product pages checked describes a technical check that a cloned voice belongs to the uploader; the custom-cloning safeguard is the Terms.
The Terms grant users a non-exclusive, irrevocable, royalty-free, worldwide license to use custom voice model output for personal and commercial use. Outputs from royalty-free library voices carry a royalty-free license for personal and commercial use, including streaming monetization, provided the user had full rights to the input. For royalty-free vocals and trained voices alike, the help center goes further in wording, saying users who have rights to all input data retain full ownership and usage rights over outputs generated on Kits; the Terms frame the same outputs as licensed.
Not without authorization. The Terms describe conversion of your own voice or one you have authority to use, and they ban impersonation without authorization, including training a model on recordings you have no authority to use. The built-in library uses licensed voices named by style rather than by artist.
Yes, within the stated conditions. Outputs from your own trained models are licensed for personal and commercial use, and library-voice outputs carry a royalty-free license, both on the condition that you held the rights to the audio you put in.
The Free card lists 0 download minutes, so downloads effectively start with a paid plan.
They are kept but frozen until you renew, while fixed download minutes are lost once the subscription expires.
For accounts created on or after October 2, 2024, it may. You can ask by email to exclude future uploads and outputs, but material provided earlier, and models already trained on it, can still be used.
