Voice Cloning · 12 Languages
The voice you build shows up in the other tools, ready to narrate podcasts, read text, and voice slides and explainers.
Record by chatting in the browser, or upload an audio file you recorded yourself — whichever is easier.
Talk naturally rather than reading flatly, in a quiet room without echo. Uploads accept wav, mp3, or m4a up to 20 MB.
Pick English, Chinese Mandarin, Japanese, Spanish, Portuguese, French, German, Turkish, Korean, Italian, Thai, or Vietnamese, then name the voice and set its gender so it is easy to find later.
Listen to the preview before saving. Once saved, the voice appears in the voice picker across the other tools.
Publish episodes that sound like you without sitting down to record each one.
Keep one consistent voice across a deck or a video series, even when you edit the script later.
Send a report through text to speech and have it read back in a familiar voice.
Use the same voice across every piece of audio you publish, instead of a different stock voice each time.
Voice cloning builds a synthetic version of a specific voice from a short recording, so that voice can then read any text you write.
About 25 to 35 seconds. Speak naturally — free conversation or energetic delivery works better than flat reading — in a quiet, low-echo room.
You can upload a wav, mp3, or m4a file up to 20 MB. Use audio you recorded yourself rather than downloaded or third-party files.
Cloning supports English, Chinese Mandarin, Japanese, Spanish, Portuguese, French, German, Turkish, Korean, Italian, Thai, and Vietnamese.
A saved voice shows up in the voice picker for AI Podcast, Text to Speech, Slides, and Explainer Video, so one clone covers all of them.
Either works. Voice Chat records you in the browser while you talk, and Upload File takes an existing recording — both produce the same kind of voice.