
nanavoice
Free AI voice cloning from a short sample, no account needed
About nanavoice
nanavoice is a browser-based AI voice cloning studio that runs on CosyVoice models. You record or upload a clean sample of 3 to 30 seconds of one person speaking, confirm that you own the voice or have written consent, then type a script of up to 2,000 characters and hear it read back in that voice. It supports 11 languages and three voice models, needs no account and is free during its open beta. Each voice profile belongs to the person who uploaded it and can be deleted. It is aimed at creators who want narration in their own voice.
Record a sample in the studio, or upload an MP3, WAV, M4A, OGG or FLAC file of 3 to 30 seconds with no music and little background noise. Tick the consent box and clone the voice; the sample goes to the speech provider to build a profile and is not used for training. Then choose a model, type up to 2,000 characters and render the audio. New voice deletes the profile upstream.
- •Short-sample cloning - Builds a voice profile from 3 to 30 seconds of clean speech recorded in the browser or uploaded as a file.
- •11 languages - Speaks English, Mandarin, Japanese, Korean, French, German, Portuguese, Russian, Thai, Indonesian and Vietnamese in the cloned voice.
- •Three CosyVoice models - Lets you choose between models, from a fast low-cost flash model to a higher-quality studio model.
- •Consent gate - Requires you to confirm you own the voice or have written consent and will not use it to impersonate or deceive anyone.
- •AIGC watermark - Marks output from the Studio model with an AI-generated content watermark, according to the site.
YouTubers and course creators use it to narrate videos in their own voice without re-recording every change. Podcasters use it to patch individual lines. People who publish in several languages use it to keep one voice across them.
Pricing
$0/mo
- Open beta$0/mo
























































