Clone Your Voice in the Browser and Use It in Every Voice Picker
Voice Cloner builds an AI copy of your voice from about 30 seconds of speech. Read the short script out loud in your browser, or upload audio and video clips and drag across the waveform to pick the stretches you want used. The clone is built from that sample and appears at the top of every voice list on Reality Fabricator.
No recording at all? Describe a voice in plain words instead: who they are, their age, their accent and how they sound, such as a gravel-voiced old Cornish pirate who is amused by everything. An AI voice designer builds it in about twenty seconds. A designed voice is nobody's real voice, it takes a direction on every line, and it is ready for Voice Scenes, where a whole cast reads a script together.
That means one recording covers a lot of ground. Your companions and adventures can speak in your voice in audio mode. So can the companion desk device, living avatars, Based AI, the household guardian, document-reader audiobooks and the narration you type into the video editor. It is meant for creators who want their own voice on everything they publish without re-recording every line.
The script includes a spoken consent sentence, and the clone is only built if that sentence was actually said. Uploaded clips require an ownership attestation instead. You manage every voice in one place: listen to it, rename it, see which companions, devices and avatars use it, and delete it.
Open Voice Cloner on Reality Fabricator
What you can do
- Record a 30-second script in the browser, no upload or software needed
- Or describe a voice in words and have it designed for you in about twenty seconds
- Or upload audio and video clips and select the stretches to use on a waveform
- Your clone appears first in every voice picker across the site
- Works for companions, the desk device, living avatars, audiobooks and video narration
- A spoken consent line is checked before any clone is built
- Hear, rename, trace usage and delete each voice from one manager
How it works
- Name the voice, then record the on-screen script or upload clips and pick the good stretches.
- The sample is checked for the consent sentence and the clone is built.
- Listen to the result, then press Use this voice to return to where you came from.
- Pick it from the top of any voice list on the site.
Frequently asked questions
How much audio do I need?
About 30 seconds. The script you read in the browser is enough. If you upload clips instead, you select stretches from them and the total must fall between 12 and 120 seconds. Only the stitched sample is sent, never your full source files.
Can I clone someone else's voice?
The recorded path requires you to speak a consent sentence that is checked before the clone is built, so it only works for a voice you own. The upload path asks you to attest that you have the rights to the voice you are cloning.
Where can I use my cloned voice?
Anywhere a voice is chosen: companions and adventures in audio mode, the companion desk device, living avatars, Based AI, the household guardian, document-reader audiobooks and typed narration in the video editor. Your voices appear in a Your voices group at the top of each picker.
How many voices can I keep?
Up to 10 per account. The manager lists each one so you can hear it, rename it, see which companions, devices and avatars use it, and delete it. If a deleted voice is still assigned somewhere, that item falls back to a stock voice rather than going silent.
Does it cost tokens?
Cloning a voice is free, and short previews are free. Speech generated with your clone is billed at the same per-character text-to-speech rate as the stock voices. Designing a voice from a description costs about 2,500 tokens, and directable voices cost about twice a stock voice per line because they are billed on the audio the model actually produces.
What is the difference between a cloned and a designed voice?
A clone copies a real voice, yours, from a recording, and only works with your spoken consent or an ownership attestation. A designed voice is invented from a written description, so it belongs to no real person. Designed voices also take a direction on every line, such as whispering, panicked or slow, and can be shared: the players of a story hear the voices its author designed.