AI Voice Cloning
Can't find the voice you need? Clone one from a short audio or video clip, then turn any text into speech in that voice.
Upload a voice sample
An audio or video file with one person speaking clearly. 5–60 seconds works best; longer files use the first 60 seconds. Avoid background music.
How voice cloning works
- 1
Upload a clip
Pick an audio or video file with one person speaking clearly. 5 to 60 seconds is enough.
- 2
We build the voice
We transcribe the clip and create a voice model from it. It takes about 30 seconds.
- 3
Type and generate
Your voice appears under My voices in the voice library. Type any text and generate speech with it.
Voice cloning FAQ
What kind of recording works best?
One speaker, clear speech, no background music, and at least 10 seconds of talking. Phone recordings are fine; noisy or echoey rooms lower the quality.
Can I upload a video?
Yes. MP4, MOV and WEBM files work. We extract the audio in your browser and only upload the sound, never the video.
Does cloning count against my generations?
Creating a voice doesn't. Generating speech with it counts the same as any other voice, including the 2 free generations per day on the free plan.
Who can use my cloned voice?
Only you. Cloned voices are private to your account, and you can delete one at any time, which also deletes the uploaded sample.
Can I clone someone else's voice?
Only with their permission. Cloning a voice to impersonate someone, deceive people or break the law is against our Terms, and we remove voices reported for misuse.