Record your consent and a short sample of your own speech. Save a private voice, then use it to read new English scripts in Studio. Listen to a first take before deciding whether the voice and pronunciation suit your narration.
One output, up to 300 characters, for evaluation. Verify your email. No card required.
After signing up and verifying email, open Clone voice. Allow microphone access and read the statement shown there live. Your consent recording is separate from the voice sample and is required even when you upload a sample file.
Record 15 to 30 seconds of natural English speech, or upload your own WAV, MP3, M4A or WebM recording. Keep one speaker, no music and no clipped peaks. The sample file must be under 6 MB.
If you add a sample transcript, copy the recording word for word and confirm it matches. Give the voice a name and save it to your private library. The transcript is optional; do not confirm words that differ from the audio.
Use your saved voice in Studio with a new script. Listen for your vocal character, the intended words, product names and the ending. Keep the same saved voice for revisions; each successful new take uses its own allowance.
Aim for a quiet recording, not a polished performance. A clone is a generated interpretation of your sample; identical sound and pronunciation are not guaranteed.
Speak naturally in a quiet room at a steady microphone distance. Aim for 15–30 seconds; the app rejects samples under 3 seconds or over 2 minutes, and warns about very short, long or clipped recordings. Check the sample yourself before saving.
Use a different sentence for the first generated take. Compare it with your reference for vocal character and pronunciation. If it is not useful, check background noise, clipped audio and transcript accuracy before recording a cleaner sample.
Cloning is for your own voice. Do not submit another person’s recording or use a voice to impersonate someone. Your saved clone is private to your account. Deleting it removes its reference and consent recordings and associated generated audio; download anything you need to keep first.
Privacy PolicyVerify your email to try one output of up to 300 characters, for evaluation only and with no card. Stock voice previews are free. A successful voice design preview uses the free output. Paid plans and character packs support publishing audio you have the rights to use, subject to the Terms. Disclose synthetic audio when required.
Paid offers allow up to 5,000 characters per request. Every successful revision uses its own allowance; failed or cancelled jobs release their reservations. Playback and WAV downloads are free.
Yes, you can upload your own WAV, MP3, M4A or WebM sample under 6 MB. You still need to record the displayed consent statement live with your microphone in the app.
A saved clone generates new speech from your sample; it does not replay your recording. Voice similarity and pronunciation vary with the sample and script. Test a short new sentence, compare it with your reference, and listen before publishing.
After email verification, you can save one private voice and generate one output of up to 300 characters for evaluation only, without a card. Deleting a voice frees its slot but does not reset the output allowance. A successful design preview uses the same free output.
The free sample is for evaluation. With a paid plan or character pack, you can publish audio that you have the rights to use, subject to the Terms. Clone only your own voice.