Create a new narration from your own voice
Instant voice cloning uses a reference recording to shape new speech. Upload only your own voice or a recording you have explicit permission to clone. Do not impersonate another person or present synthetic speech as an authentic recording.
How to use this tool
Record your voice in the browser or upload a clean 10–60 second MP3 or PCM WAV with one speaker. Name your voice and confirm permission. Studio transcribes the recording automatically, then saves the reference privately for Text to Speech. You confirm the credit price before preparation starts.
Tips for a better result
- Record one permitted speaker in a quiet room. Use natural speech at a consistent distance from the microphone, without music, effects or long silences.
- Give the voice a recognisable name so you can find it later. Prepare a clean reference recording within the displayed duration and file limits.
- After preparation, select the saved voice in Text to Speech and try a short script. Review pronunciation and identify synthetic narration when sharing it.
Frequently asked questions
Does this create a public or professional voice model?
No public model or professional training is created. Each generation sends the reference to Fish Audio. If you choose Save voice, Studio encrypts your reference and transcript for reuse for 30 days, with up to five saved voices and deletion available here. Unsaved results expire after seven days. Save a recording to protect it and its reusable voice from automatic cleanup.
Membership, credits and privacy
You can explore the tool before signing in. Generating requires an active Studio membership and enough credits. Eligible new members can start with 50 credits for $0.99 for seven days, then $9.99/month with 100 credits unless cancelled. Review the current terms when joining; the credit price for your recording or edit is shown before it starts.
Results are private to your account. Unsaved results expire after seven days. Use Save to keep results in your Saved section and protect them from automatic cleanup. Voice references remain available for 30 days, or longer while their recording is saved. You can delete results and voices at any time.