Kits.ai
AI voice cloning and voice conversion for music: train a model of a singing voice, swap vocals, and clean up or split recordings in the browser
Decision snapshot
Fast-read signals for fit, pricing, and trust.
- Category
- Video & Audio Production
- Best for
- ai voice cloning
What is this tool?
Most useful to producers and songwriters who want demo or backing vocals in their own trained voice, or in a licensed library voice, without booking a singer. Test your own voice before relying on it, because results depend on a clean, dry recording and users report mixed quality. Read the terms too: they give Kits ownership of the voice models you train and, by default, the right to train its models on your uploads, with an email opt-out that applies only from then on.
By QT Desk
Read full overviewHide full overview
Kits is organised around voices. You either train a professional model on a dataset of your own clean vocals or make an instant clone from a short clip, and the model sits in a voice slot in your account. From there you work in Kits Studio in the browser: upload or record a vocal, pick the voice, convert, and download the result if your plan has download minutes. The same studio holds the clean-up tools, the vocal remover and stem splitter, and other tools named on the homepage such as AI mastering and harmonies. A Windows desktop app lets you drag converted audio straight into your DAW; it is an app, not a plugin, and Mac availability is not confirmed on the download page.
Conversion minutes cover turning audio into another voice, and download minutes cover taking the result out. Free has 15 conversion minutes a month, Starter adds cloning and 15 download minutes, Producer raises downloads to 60 minutes, and Professional removes the download limit, with "unlimited" always subject to fair use. Prices are for new accounts; annual billing is offered at up to 47% off, but the annual totals are not shown on the plan cards. Kits' help centre says the API is open to Starter, Producer and Professional users, while the API page asks you to request a beta key.
The terms deserve a careful read before you clone your own voice. You may only clone your own voice or one you have authority to use, impersonation is prohibited, and users must be 18 or older. Kits owns the voice models in its library and the custom models users train, and gives you a licence to use your custom model. For accounts created from October 2, 2024, Kits may train its models on the voice files you provide and the output you generate, unless you opt out by email, which covers only what you provide afterwards. Your voices are private to you by default, which keeps them from other users but does not change that training licence.
Is this right for you?
Quick fit check against how you actually work.
Choose this if
- You want scratch or demo vocals in your own voice and can record 10 or more minutes of clean, dry singing to train on
- You need a royalty-free singing voice for a release and want the library's artist-licensing terms rather than a fan-made model
- Noisy or reverberant vocal takes need cleaning up, and vocal removal and stem splitting belong in the same workflow
- You work on Windows and want converted vocals dragged straight into your DAW
Consider alternatives if
- Free downloads are essential to you, since the Free plan includes no download minutes
- Owning the model of your own voice matters to you, since Kits' terms say custom models belong to Kits
- Your recordings cannot be cleaned up to dry, single-voice takes, which is what Kits and its users say good clones depend on
- A native VST or AU plugin, or a confirmed Mac app, is a requirement for your setup
- You plan to clone a singer whose permission you do not have; the terms prohibit it
- Your main use is swapping the vocal on finished AI-generated songs, where users report glitches or cut words
What people say
Aggregate community sentiment, summarized — not native reviews.
Mixed signals
Reports on clone quality are mixed. Several users say their clone came out robotic, distorted or unlike their own voice, while others say the similarity was convincing. Users who were happy with the result point to clean, dry recordings as the reason. Users replacing the vocal on Suno songs report glitches, cut words or robotic sections, and some tie it to backing vocals or instruments left in the track.
Key features
What you can actually do with this tool.
Professional voice cloning
Upload or record dry, monophonic vocals, press Train and wait from about half an hour to several hours. Kits recommends 10+ minutes of vocals, and its help centre suggests 15-30 minutes or more; the cleaner the dataset, the closer the model gets to the voice.
Instant cloning when you need a quick sketch
An instant clone is built from roughly 30 seconds of dry vocals with no training wait. Kits itself says it is less similar to the target than a trained model and that results "might not sound exactly like your dataset", so treat it as a draft tool.
Voice conversion with a licensed library
Convert an uploaded or recorded vocal, or a YouTube link, into your own model or one of the library voices (Kits says 100+). Kits says the library voices are licensed from consenting artists who share in the revenue, that its models are Fairly Trained certified, and that library output is cleared for commercial use provided you have rights to the audio you upload; each voice page carries its own terms.
Vocal clean-up and repair
AI Vocal Repair restores clarity and removes background noise from sung or spoken recordings, Vocal Enhance removes noise and boosts clarity, and the vocal remover can strip noise, reverb and echo. Kits notes Vocal Repair may do less on vocals that are already mixed or processed.
Vocal removal and stem splitting
Separate vocals, instrumental and backing vocals from a file or YouTube link, or split a track into vocals, drums, bass and instruments. This is also the step to run before swapping the vocal on a finished song.
Plans & pricing
Pricing tiers and what's included in each.
Free
$0 a month
- 15 conversion minutes a month and the voice designer and blender, with no credit card required
- No download minutes, so audio cannot be downloaded, and no instant or professional voice cloning
- Kits' pricing page and help centre give different voice-slot counts for this plan
Starter
$10 a month billed monthly (US dollars); a discounted annual option is offered
- Instant and professional voice cloning, 2 voice slots and 15 download minutes a month; Kits' help centre says unused download minutes roll over
- Advanced settings and the choir tool; the pricing page and help centre disagree on how many conversion minutes are included
Producer
$30 a month billed monthly (US dollars); a discounted annual option is offered
- Unlimited conversion minutes and voice slots, subject to fair use, and 60 download minutes a month
- High-fidelity model quality
Professional
$60 a month billed monthly (US dollars); a discounted annual option is offered
- Unlimited conversion minutes, voice slots and download minutes, subject to fair use
Official pricing pagePricing may have changed. Always verify before purchasing.
Frequently asked
Common questions about this tool, answered.
How much does Kits AI cost?
Billed monthly in US dollars, Starter is $10, Producer $30 and Professional $60 a month, and the Free plan costs nothing. Kits also offers annual billing at up to 47% off but does not show the annual totals on its plan cards. Prices are for new accounts. Plans renew automatically: you have to cancel three days before the renewal date to avoid the next charge, and payments are non-refundable, according to the terms.
Can I use the free plan to make songs?
Only to try the tools. The Free plan gives 15 conversion minutes a month and the voice designer, but no download minutes and no voice cloning, so getting audio out of Kits and training your own voice both need a paid plan, starting with Starter.
Can I release music made with Kits commercially?
Kits says library-voice output is cleared for commercial use provided you have rights to the audio you upload, and each library voice has its own terms to check. For models you train yourself, the terms grant you a licence to use the output for personal and commercial use, including on streaming platforms. Kits' help centre and its terms describe your rights in the output differently, so read the terms rather than assuming you own the output outright.
Does Kits use my voice to train its AI?
By default, yes, for accounts created from October 2, 2024: the terms and the privacy policy let Kits train its models on the voice files you provide and the output you generate. You can opt out by emailing Kits, but the opt-out applies only to what you provide afterwards.
What happens to my voices if I cancel?
Your custom voices are not deleted; they are locked and unavailable until you subscribe again. You keep access to paid features until the end of the period you have paid for.