Clone your own voice

Read a paragraph. Give your voice a new sentence. Explore what the model preserves and what it changes.

How this experiment works

A pretrained speech model encodes your reference recording and conditions new speech on it. This is inference, not training a speech model from scratch.

What it does not prove

A clone is an approximation, not identity verification. Use your own voice. Audio stays in this tab; generated speech is labelled as synthetic.

Try this challenge

Generate the same sentence from a clean and a noisy recording. Compare pronunciation and likeness.

Enable JavaScript to run the experiment on your device.