Start a Flash clone from a clean 10-second sample, then convert your recorded speech or singing through it. Your voice model stays private. Try Flash with free credits, or choose Pro for a longer training workflow.
Free tier · 30 credits / month · Voice models always private.
Start with a clean recording of speech or singing and a target voice you have permission to use. This workflow changes an existing recording; it does not generate speech from typed text or change a live microphone feed.
A Flash clone costs 10 credits. A 30-second conversion costs 1 credit; a 3-minute conversion costs 3 credits. Each conversion rounds up to whole minutes. Creating one Flash clone and converting those two recordings costs 14 credits in total. Separation, mastering, and other operations cost extra. Check the quote before submitting.
The free plan includes 30 credits per month. Paid plans provide more credits and voice slots, Pro cloning, and commercial use. Compare the current plan details before upgrading.
Record close to the microphone in a quiet room. Avoid clipping, reverb, and background music in your reference. Start with a short phrase, listen to the result, and adjust your recording or target voice before converting a longer take. Quality and processing time depend on the source, target voice, and workload.
For a meme, record an original joke in your own character voice. Label synthetic parody when sharing it, and keep the original recording alongside the conversion so listeners can hear the difference.
Same canvas, two reference budgets. Start with Flash, upgrade to Pro once a demo is worth shipping.
Record or upload a single clean clip. A phone mic is fine.
Ideation, demos, social video, anything you want to test before committing to a release.
A curated dataset covering your full pitch range, recorded in a quiet room with a consistent mic.
Releases, character voices, language redubs, anything where the result has to hold up at full mix loudness.
A 10-second clip for Flash, 5 minutes of clean reference for Pro. Phone mic works for Flash; Pro wants a quiet room and consistent levels. Read a sentence, sing a line, or both.
Use the clone you just made, or pick from the voices already in your library. Every voice is private by default. Your trained models never leave your account and never get listed publicly.
Drop any vocal card on the canvas and convert it through the chosen voice. Conversion aims to preserve your phrasing while changing the voice. Preview the result and check your plan for export formats.
Record yourself in your home setup, then convert the vocal through any voice in your library. Phrasing and breath stay yours. Only the timbre changes.
Producers use voice cloning to demo a topline in the target artist’s voice before sending it over. Faster than booking a session and a recorded performance to compare with the original.
Re-record the corrected line, then convert it through a clone of the host. Compare the patch with the surrounding recording before editing it in. This workflow uses recorded audio, not typed text.
Re-record the localized script in any voice, then convert through your talent’s clone. The result keeps the original speaker’s identity in a language they don’t actually speak.
Build a roster of character voices from short reference clips. Iterate on dialogue without scheduling actors, then layer a real performance for the hero scenes.
Layer three or four versions of yourself with different pitch shifts to fill out a chorus. Cheaper than booking a session singer, and the texture stays consistent.
Flash starts from a short reference for quick experiments. Pro uses a longer dataset to train a reusable voice. Processing time varies with the dataset and current workload.
Listen to each converted recording before downloading it. Pricing lists the available export formats and commercial-use permissions for each plan.
Voice models live in your account only. There is no public model marketplace, no listings, no opt-in sharing toggle that flips by accident. Your voice is not training data.
Every Pro voice training run requires an explicit consent attestation. You cannot upload someone else’s voice and skip the checkbox. That audit trail protects both you and the talent.
Voice cloning is one card on the same canvas as your generation, stem split, and mastering. No tab juggling between five different tools to finish a track.
Credits, not seats. You pay for what you convert. The free tier ships with enough credits to clone, convert, and export your first track without typing in a card.
Flash starts from a clean 10-second sample; Pro trains from a longer, varied dataset, often around 5 minutes. Processing time depends on the sample, training settings, and workload. Check the task status after submission.
10 seconds of clean speech or singing is enough for a Flash clone. For a Pro clone, plan on 5 minutes of high-quality reference covering your full pitch range. The cleaner the source, the cleaner the result. Phone recordings work for Flash, but Pro really wants a quiet room.
No. Every voice model you train is private to your account. There is no public voice marketplace and no opt-in flag that can flip the model into a public listing. Voices belong to you and stay with you.
You own the audio you generate. Commercial use is allowed on paid plans. If you are cloning someone else’s voice for a release, the responsibility for clearing rights with the original speaker sits with you. The consent attestation step exists to make that explicit.
Upload MP3, WAV, FLAC, or M4A. Check the pricing page for plan-specific export formats. Preview a short conversion before committing to a longer recording.
Yes. The free plan ships with 30 credits per month, enough to clone a voice, run a few conversions, and export a finished track. No credit card required to start.
30 free credits. No card. Your voices stay private, always.
