Cloning Your Voice
Create a private reusable voice from a clean recording or audio file.
Create a private cloned voice
Voice cloning creates a reusable TTS voice from a recording. Only clone a voice you own, have consent to use, or are otherwise legally allowed to use.
Open My Voices, click Build New Voice, and choose Instant Voice Clone.
1. Record or upload a sample
You can record in the browser or upload an MP3, WAV, M4A, or WebM file. The interface requires at least 10 seconds; around 30 seconds of good material is enough for most voices.
For a stronger result:
- Record one speaker only.
- Use a quiet room with little echo.
- Avoid music, sound effects, and aggressive noise removal.
- Speak naturally at a steady distance from the microphone.
- Include varied sentences rather than repeating one phrase.
Preview the uploaded clip and replace it if the speech is clipped, distorted, or hard to understand.
2. Analyze and describe the voice
Click Start analyze, then complete Voice Details:
- Give the voice a recognizable name.
- Add a short description.
- Choose useful metadata such as language, gender, age, use case, and style.
- Confirm that you have permission to create and use the voice.
Click Create. New voices remain private in My Voices; publishing and sharing are not currently enabled.
3. Use the cloned voice
From My Voices, click Use in TTS, or open the voice picker in TTS Studio and select the My Voices tab.
Custom voices count toward the voice-slot allowance shown on the My Voices page. Delete an unused custom voice if you need to free a slot.
Voice cloning copies vocal characteristics. It does not automatically fix background noise, pronunciation, or unclear delivery in the source sample.
Prefer an original voice?
Choose Voice Design under Build New Voice to describe a new voice in text, generate a sample, and save it without cloning a real speaker.
Related guides
Put the guide into practice
Start with your own document and keep the script editable from source to speech.