Skip to content

OpenVoice V2

Clone a voice across languages with a TTS model

Visit official site

What OpenVoice V2 is

The model is described as a text-to-speech model released in April 2024 that keeps all V1 functions and adds improvements, with a new training strategy and better audio quality.

What you can do with it

  • Generate speech in English, Spanish, French, Chinese, Japanese and Korean
  • Clone the tone color of a reference voice
  • Control emotion, accent, rhythm, pauses and intonation
  • Clone voices across languages with zero-shot cross-lingual cloning

Application scenarios

  • Give video game characters realistic voices
  • Generate teaching content for learners of different languages
  • Produce multilingual narration for commercial ads

Getting started

  • Visit the product page
  • Choose quick use or a local install
  • Try deployed services such as British or American English
  • Clone the repository and follow the guide on Linux

Pros & cons

✓ What we like

  • Free commercial use under the MIT license since April 2024
  • Style controls cover emotion, accent, rhythm, pauses and intonation

! What to watch out for

  • Tone color cloning starts from a reference voice
  • The Linux path requires cloning the repository and following the guide

FAQ

Which languages are supported?

The feature list names English, Spanish, French, Chinese, Japanese and Korean.

Can it be used commercially?

The material states V2 and V1 are released under the MIT license since April 2024, allowing free commercial use.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review