/Voice-Cloning-App

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

Voice Cloning App

A Python/Pytorch project for easily synthesising human voices.

Preview

Key features

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

System Requirements

  • Windows or Linux operating system
  • NVIDIA GPU with at least 4GB of memory

Video guide

https://www.youtube.com/watch?v=ccvjGKiPenQ&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

  • Test pretrained weights for transfer learning
  • Add support for alternative models