Generate new speech from a script using a reference voice recording.
Get the right result: check these first
Recording and script do different jobs. The recording supplies the voice reference; Enter what you want the person to say supplies the new words. Use Voice Changer if you want to start from an existing spoken performance instead.
Microphone blocked. You can upload an audio file using the upload area instead of recording in the browser. A microphone recording is not the only input method.
Missing language choice. Use the language choices actually shown. Auto is not a promise that an unlisted language is supported; confirm support before relying on it for your project.
Still stuck? Contact support with the tool name, project link or visible ID if available, reference recording, script, selected language and speed, and timestamp of any missing words or timing change. Keep the project while it is reviewed. How to send useful troubleshooting details.
Step by step
Open Voice Cloner and sign in.
Upload a reference recording or select Record and allow microphone access. The recorder recommends 10–30 seconds for best results; this is guidance, not a universal file limit.
Enter the new script in the text area.
Choose the language and Speed, check the reference and estimate, then select Generate Speech.
Listen to the completed speech and download it. Check names, omissions, and pronunciation before using a long passage.
Use this recording now, or save it for reuse
A new upload or recording is the current reference; it is not automatically a saved My Voices item. To keep a supported uploaded or recorded reference for later, use the bookmark control labelled Save voice, name it, and select Save to Library. Confirm that it saved before relying on it in another session.
Selecting a preset or a My Voices item replaces the active reference. Before Generate Speech, play or check the current reference and confirm it is the intended voice. Saving a voice and selecting it are separate actions.
If the preview is correct but generated speech repeatedly uses another voice, stop paid retries and contact support with the tool, project link, and selected-reference screenshot. Do not assume switching to Voice Generator will fix a wrong-reference problem.
Remove a current voice reference versus delete a saved voice
In a video tool’s Add Audio picker, choose Clone your voice with AI. After uploading or recording a reference and reaching the script step, open Change voice (Change on mobile), then Remove current voice, if that control is shown. This clears the current reference recording and the draft script, so copy any script you want to keep first. You can then upload or record the intended reference. If that control is not shown for your current input, contact support with the tool name and a screenshot; do not assume the saved voice has been deleted.
Clearing this draft is not deletion of a saved voice in My Voices or of audio already generated. If you want a saved voice removed and cannot find an applicable delete control, contact support with the saved item’s name and the tool or screen where it appears. Do not assume it has been permanently deleted because the current recording disappeared, or delete your account merely to replace a reference.
A useful reference recording
Use one clearly audible speaker with minimal music, echo, or competing speech. A noisy or mixed-speaker recording makes it harder to identify the intended voice.
Cloning versus changing a recording
Voice Cloner creates new speech from text. It does not promise to preserve the reference recording’s exact words, timing, pauses, or emotion. Use Voice Changer when the input is an existing performance you want in another voice.
Wrong language or cut-off words
Choose the language that matches the script; the language control does not translate text. Try a short passage to check pronunciation. For repeated missing words, send support the project link, script, language, and timestamp.
Cost, limits, and your result
These instructions cover the Dashboard. Sign in to generate and check the displayed credit estimate and access requirements. File limits and settings can depend on the tool, model, and plan. Follow the upload and validation messages shown for your selection.
Review the completed result before downloading. For signed-in projects, see find and download your creations.
If a job is still processing or failed
For a job still processing, check its status before submitting again. For a failed or technically broken output, see generation failures and credits.