AI Voice Cloner
Record ten seconds of your voice, paste a script, and download a natural voice-over in your own voice. Everything runs in your browser.
How it works
Start the engine
Pick a model size. It downloads once and is cached, so returning visits load in seconds.
Record your voice
Read the short prompt in a quiet room. Your voice is saved in your browser, you never redo it.
Paste your script
Write anything. Long scripts are split by sentence and stitched into one smooth take.
Download
Preview the result, then save it as an MP3 for your videos, ads, and social posts.
Tips for a cleaner clone
- Record in a quiet, soft room. Curtains and carpet beat an empty kitchen.
- Speak the way you want the voice-over to sound. The clone copies your energy.
- Use a headset or phone mic close to your mouth, not a laptop mic across the room.
- Write scripts with normal punctuation. Commas and full stops create natural pauses.
- Use the expressiveness slider: lower for calm explainers, higher for ads and hooks.
- Generate long scripts in sections if one paragraph sounds off, then regenerate only that part.
Frequently asked questions
Is my voice uploaded anywhere?
No. The model downloads into your browser and everything runs on your device. Your recording, your script, and the generated audio never touch a server. Saved voices live in your browser's own storage.
How much audio do I need to record?
Eight to ten seconds of clean, natural speech is enough. Record in a quiet room, keep the mic a hand's width from your mouth, and read at your normal pace. You only do this once, your voice is saved for next time.
How big is the download and do I need a good computer?
The Standard model is about 1.5 GB and the High quality model about 2.2 GB. Both download once and are cached. A desktop with a dedicated GPU in Chrome or Edge is fastest. The tool checks your device and recommends a model, and it warns you when your hardware will be slow.
Why does a long script take a while?
The script is split into sentences, each one is generated on your GPU, and then everything is stitched together. Generation speed depends on your graphics card, so a laptop GPU is slower than a desktop one.
Do I have to clone my own voice?
No. Open the Templates tab and pick one of the ready-made voices, male or female, then write your script. Template clips load from a CDN only when you choose one, so nothing extra is downloaded up front. They come from LibriTTS-R (CC BY 4.0), based on public domain LibriVox recordings.
Which languages are supported?
This version is tuned for English. The underlying model is Chatterbox by Resemble AI, released under the MIT license.
Can I use the audio commercially?
The model license allows it. Only clone your own voice, or a voice you have clear permission to use. Cloning someone else's voice without consent can be illegal.
Need voice-overs built into your product?
I build custom voice, audio, and in-browser AI tools for brands and creators, from branded voice-over generators to private pipelines that never send your data to a third party.
Get a Custom Quote