AI voice cloning
Make new speech with a voice you can reuse
Create a voice from a clear recording of yourself or a speaker who has given you permission, then use it to read new text. My Voices keeps saved voices and speech controls together. Try a short script, listen to the result and use it in a talking photo, a saved avatar or video narration.
Available models, settings and costs are shown inside the product before you create.
Welcome to the studio. Today I will show you how this little clay bowl takes shape, starting with the piece of clay on the table.
How it works
Record, save and try your voice
- 01
Prepare the voice sample
Record 10 to 60 seconds of one person speaking clearly in a quiet place. Use their natural speaking voice and keep music and other speakers out of the sample. Check the file requirements in the form and confirm that you have permission to create and use the voice.
- 02
Create and name the voice
Add the recording in My Voices, choose its language and give the voice a name you will recognize later. When creation finishes, the voice appears in your collection. You can return to the same voice for another script instead of supplying a new recording each time.
- 03
Test a short piece of speech
Write a few representative sentences and choose the available speech settings. Review the credit estimate, generate the audio and listen. Include a name or phrase you expect to use often. Once the reading suits the project, you can download it or continue into a talking video.
Practical guidance
A good sample gives the voice a better starting point
The recording establishes how the speaker sounds. Spend a little time getting a clean sample, then judge the generated speech with words you actually intend to use rather than a generic test sentence.
Keep the recording consistent
Speak at a comfortable distance from the microphone and stay there through the sample. Avoid switching between whispering and shouting. If a fan, traffic or room echo competes with your voice, choose a quieter recording spot and make another sample before creating the clone.
Test the words that matter
Include the product name, place name or personal name that will appear in the finished piece. Listen for pronunciation and sentence rhythm as well as resemblance. A clone interprets the sample, so the result can vary with the script. A convincing tone does not mean every unusual word will be read the way you expect.
Give punctuation a purpose
Short sentences and clear punctuation make a script easier to hear. Read it aloud yourself before generating. If you run out of breath halfway through a sentence, divide the thought. Start at normal speed, then adjust the pace in any supported language. Emotion controls are available in English only.
Try another language carefully
Choose a supported language for the script you want spoken. My Voices can also add a native accent for the languages offered by that voice control. Listen to the result with someone who knows the language when the pronunciation matters to your audience. Adding a language has its own displayed cost.
Keep your voice collection understandable
Use names that explain which recording a voice came from, such as studio voice or calm narration. You can rename or delete a saved voice in My Voices. A saved avatar can use a voice from the same account, so recognizable names also help when choosing a speaker for a video.
Questions
Frequently asked questions
Whose voice can I clone?
Use your own voice or a recording from a speaker who has given you permission to clone and use it. The creation form asks you to confirm that permission. Keep the intended script and use within what the speaker agreed to, including any video made with their voice.
How long does the sample need to be?
Prepare 10 to 60 seconds of clean, continuous speech from one speaker. You can record in the form or upload an audio file, then check the sample before creating the voice. Background sound or several voices makes it harder to learn the intended speaker.
Can the same voice read different scripts?
Yes. Once the voice is saved, select it in My Voices and write a new script. You can listen to each generated reading and download the audio. Speech has a credit estimate before generation, so check it after editing the text.
Can I use the voice with my photo?
Yes. Generate the speech, then use it with a photo in Lip Sync. You can also attach a saved voice to an avatar so that its photos and voice stay together. Speech creation and talking video creation are separate steps you can review.
Start with the idea you have now
Bring a clean recording from a speaker who has agreed to its use. Create the voice, then try a short script and listen before making your full piece.
Create your voice