/Voice-Cloning-App

A Python/Pytorch app for easily synthesising human voices

Primary LanguageJupyter NotebookBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

Voice Cloning App

CircleCI

A Python/Pytorch app for easily synthesising human voices

Preview

System Requirements

  • Windows 10 or Ubuntu 20.04+ operating system
  • NVIDIA GPU with at least 4GB of memory
  • Up-to-date NVIDIA driver (version 450.36+)

Key features

  • Automatic dataset generation
  • Easy train start/stop
  • Support for kindle & audible as data sources
  • Data importing/exporting
  • Simplified training & synthesis
  • Word replacement suggestion
  • Windows & Linux support

Video guide

https://www.youtube.com/playlist?list=PLk5I7EvFL13GjBIDorh5yE1SaPGRG-i2l

Voice Sharing Hub

https://voice-sharing-hub.herokuapp.com/

Manual Guides

  1. Installation
  2. Building the dataset
  3. Training
  4. Synthesis

Future Improvements

  • Add support for alternative models
  • Improved batch size estimation
  • Multi-GPU support
  • AMD GPU support
  • Additional language support

Experimental/In-Development features

Available for those with a manual install

Other resources

Acknowledgements

This project uses a reworked version of Tacotron2 & Waveglow. All rights for belong to NVIDIA and follow the requirements of their BSD-3 licence.

Additionally, the project uses DSAlign, Silero & hifi-gan.

Thank you to Dr. John Bustard at Queen's University Belfast for his support throughout the project.

Supported by uberduck.ai, reach out to them for live model hosting.

Also a big thanks to the members of the VocalSynthesis subreddit for their feedback.

Finally thank you to everyone raising issues and contributing to the project.