Kits AI

Free is a listening plan. The pricing table gives 15 conversion minutes a month and 0 download minutes, and the download minutes article confirms that free accounts cannot download. You can upload a vocal, convert it into library voices and hear the result in the browser, but exporting it for a session needs Starter or higher.

Voice slots are less clear. The pricing table lists 0 for Free, while the Voice Designer article says free users can create one designed voice, and a September 2026 plans article mentions one free voice slot. Treat a custom voice on Free as uncertain. A free alternative that runs entirely on your own computer is Applio, open-source voice conversion with no account, although its docs recommend an Nvidia RTX 20-series GPU or newer for training.

Kits meters two things. Conversion minutes cover the audio you render while experimenting; download minutes cover the length of audio you export, and the Harmony Generator article counts harmony layers the same way. On Starter, trying four voices per take means 60 conversion minutes cover 15 minutes of source audio, the same amount you may download. Producer removes the conversion cap, subject to fair use, and raises downloads to 60 minutes.

According to the cancellation article, unused download minutes roll over month to month, but every banked minute disappears when the subscription expires and does not come back if you resubscribe. Spend banked minutes before cancelling.

The cloning comparison describes Instant Voice Cloning as working from about 30 seconds of reference audio and ready at once, with results that may not sound exactly like the source. Professional Voice Cloning recommends 10 to 60 minutes of dry, monophonic vocals and takes from 30 minutes to several hours to train. Both are on Starter and above.

Consent is the user’s responsibility. The terms forbid impersonating anyone with a voice file you have no authority to use, and the Voice Changer article repeats that you must own the rights to all training data. The ownership check Kits describes, comparing your live vocal with the trained one, applies to Kits Earn submissions for the public library, not to every private clone.

Outputs and models are treated differently. The rights article says that, as long as you owned the input, you may use outputs of library voices and of your own trained models for personal and commercial work, including streaming releases. The terms, however, state that custom AI voice models are owned exclusively by the company and assign any rights you hold in them to Kits.

For accounts created on or after October 2, 2024, the terms also let Kits train its models on your uploaded voice files and outputs. You can opt out by emailing outreach@kits.ai from your account address, but the opt-out covers only future files; earlier uploads and models already trained may still be used. Voices stay private by default and can be shared with named email addresses.

1. Sign up on Free and note the 15 conversion minutes.
2. Upload a 30-second dry vocal you recorded yourself as WAV.
3. Convert it into three library voices and listen for timing, pitch and artifacts.
4. Check how many conversion minutes were used and that the Download button asks for an upgrade.

Accept if three conversions use about 1.5 minutes and keep the original phrasing. Reject if the meter charges more than the audio length converted. No result is recorded here.