Skip to content
Get packs

Guide · Open source

How to use OpenVoice: clone a voice and generate speech

Updated · written with AI from the project's own documentation and checked against our catalogue data

OpenVoice is an open-source voice cloning model released by MIT and MyShell. You give it a short reference recording of a voice and the text you want spoken, and it generates speech that tries to match that voice's tone color. This guide explains what OpenVoice does, what you need before starting, how to get the code, and where to find the detailed usage instructions.

OpenVoicemyshell-ai/OpenVoice · ★ 38.1k · MIT

Quick start

  1. 1Check Python is installed. Confirm you have Python on your machine, since it is the stated prerequisite.
  2. 2Clone the repository. Run git clone https://github.com/myshell-ai/OpenVoice in your chosen folder.
  3. 3Open the usage docs. Read docs/USAGE.md in the cloned repository for the build and run instructions.
  4. 4Check the QA file. Look at docs/QA.md for common issues before troubleshooting on your own.
  5. 5Prepare a reference recording. Use a short recording of the voice you want to match, with the speaker's consent.
  6. 6Generate speech. Follow the usage guide to pass the reference audio and your text into the model.

What OpenVoice does

OpenVoice takes a short reference recording of a speaker and text to say, then generates speech that tries to match the reference voice's tone color. The README describes control over style such as emotion, accent, rhythm, pauses and intonation.

The project's README describes three main advantages: accurate tone color cloning, flexible voice style control, and zero-shot cross-lingual voice cloning, meaning neither the language of the generated speech nor the language of the reference speech needs to appear in the training dataset.

  • Clone the tone color of a reference voice
  • Adjust style parameters such as emotion, accent, rhythm and pauses
  • Generate speech across languages

What you need before you start

OpenVoice is installed by cloning the repository and building it, so you need Python and basic comfort with the command line and git. The README does not describe a graphical installer or an end-user app.

Check the README and the linked usage docs for the exact environment and dependencies before you begin.

  • Python installed on your machine
  • Git, to clone the repository
  • Basic command-line familiarity

Clone the OpenVoice repository

The README's first install step is to clone the repository from GitHub. Run the command below in the folder where you want the project to live.

After cloning, open the project folder and follow the build steps in the usage documentation.

git clone https://github.com/myshell-ai/OpenVoice

Find the usage instructions

The README's How to Use section points to docs/USAGE.md for detailed instructions on running the model. Common questions are collected in docs/QA.md, which the maintainers say they update regularly.

If you get stuck, read those two files before opening an issue, since they cover the setup steps the README summarises.

  • docs/USAGE.md: detailed usage instructions
  • docs/QA.md: common questions and answers

Choose between V1 and V2

OpenVoice V2 includes all features in V1 and adds better audio quality through a different training strategy. V2 natively supports English, Spanish, French, Chinese, Japanese and Korean.

Both V1 and V2 are released under the MIT License, so for new projects V2 is the version the README highlights.

  • V2: better audio quality, native support for six languages
  • V1: the original release, with the same core features

What people use it for

  • Developers adding voice cloning or text-to-speech features to their own projects
  • Researchers studying zero-shot voice cloning and cross-lingual speech generation
  • Creators experimenting with synthetic voices in several languages
  • Teams that need a permissively licensed (MIT) voice model

Frequently asked questions

Is OpenVoice free to use?

Yes. The README states that OpenVoice V1 and V2 are MIT Licensed and free for both commercial and research use. Check the LICENSE file in the repository for the full terms.

What do I need to run OpenVoice?

The README lists Python as the prerequisite and installs the project by cloning the repository and building it. Detailed steps are in docs/USAGE.md.

Which languages does OpenVoice support?

OpenVoice V2 natively supports English, Spanish, French, Chinese, Japanese and Korean, according to the README. V1 is described as the original release with the same core features.

Is there an app I can use without coding?

No. The README does not describe an end-user app. It points to usage documentation for running the model, so it is aimed at people comfortable with Python.

How does OpenVoice compare to Coqui TTS or XTTS?

Coqui TTS, VITS, Tortoise TTS and XTTS are comparable projects in the same space. The README does not benchmark OpenVoice against them, so test each on your own audio before choosing.

Can I clone a voice without permission?

The README does not address this in detail. Cloning a voice without the speaker's consent raises ethical and legal concerns, so only use recordings of people who have agreed to it.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

OpenVoice suits developers, researchers and creators who want an MIT-licensed voice cloning model they can run and build on themselves. It is a poor fit for anyone wanting a ready-made app or guaranteed production-grade audio and support, since the README describes neither.

Install options & downloads