/Multi-Language-RTVC

User-friendly Multi-Language Voice Cloning Application

Primary LanguagePythonMIT LicenseMIT

GitHub code size in bytes GitHub top language GitHub GitHub GitHub GitHub GitHub GitHub GitHub GitHub GitHub GitHub GitHub

MLRTVC logo

Multi-Language-RTVC stands for Multi-Language Real Time Voice Cloning and is a Voice Cloning Tool capable of transfering speaker-specific audio features to synthesize speeches in that voice based on just a few seconds of unknown audio data.

Installation & Setup

In order for MLRTVC to work, you need Python installed on your machine. Visit https://www.python.org/downloads/ and download Python for your Operating System (OS).
In addition to core Python, some dependant packages are required. They can be found here.

For more information on...

...how to install Python,
...how to create a virtual environment using Anaconda,
...how to clone a repository using Git,
...how to setup CUDA, ...how to install PyTorch and ...how to get started with MLRTVC

for both Windows and Linux, visit our Installation & Setup Wiki.

License

This code is licensed under MIT. For more information regarding the license model or associated duties and rights, click here.

Ethics & Legal

With the exception of one's own voice, the cloning of voices may only be done with the voices of people who have expressly consented to the use of their voice for AI applications or whose statements, speeches or recordings are publicly available and not protected by copyright.
MLRTVC gives users the power and the potential to create statements that were never made in this form and thus represent fictions. This power binds the users of this software to the duty to use this technology responsibly, to always think about the consequences of their actions and to use cloning exclusively in the course of curiosity, science and learning.
If synthetically generated statements are used to blackmail other people, influence political debates or deliberately spread false statements in the real or digital world, this can lead to negative consequences for people and severe legal penalties.
We as developers and the MLRTVC community assume no liability or responsibility for the use of MLRTVC by others and will strictly refuse to cooperate with people who have negative intentions regarding the use of voice cloning.

Project History

This project was started in 2021 with the goal of inheriting Corentin Jemine's Real-Time-Voice-Cloning. The project originated from the wish of multi-language support for voice cloning models and is now maintained and enhanced by contributing volunteers. Visit the About us section to learn more about the team behind MLRTVC.

Contributing

We welcome all those interested in the project, from beginners to experts. The MLRTVC community standard is a nice, open-minded and efficient working climate. We encourage all those with ideas to take part in the project by sharing their thoughts.
There are multiple meaningful ways of contributing:

  • Developing code (new features, fixes, enhancements)
  • Writing documentation and Wiki entries
  • Raising issues (bugs, feature requests, enhancement proposals, code refacturing, etc.)
  • Providing pre-trained models
  • Participating in community tasks (code reviews, discussions, maintenance, etc.)

For transparacy reasons, we ask you to engage with this project via the official ways (issues, pull requests) to share knowledge and questions publicly. Only in cases where privacy or confidentiality is of great importance, other communication channels are accepted (email, chat, etc.).

Further information can be gained in the Contributing Guidelines.

Code of Conduct

Working together on this project, we share and defend certain values which are indispensable for an Open Source project like MLRTVC. For further information see here.

Help & Support

Documentation

Communication

Citation

If you use this software in your own work, software implementations or research, please cite by navigating to the Cite this repository dropdown menu in the right sidebar and choose your citation style.

Citing