UPDATE June 2020: This project is completed and the most up-to-date working repository can be found in the GLAM-Imperial group here: https://github.com/glam-imperial/EmotionalConversionStarGAN
A repository for my MSc project at Imperial College London.
The project is based around using neural networks to perform emotional voice conversion, which is the process of altering spoken word audio such that that the emotional tone is changed. The model architecture is heavily inspired by the StarGAN architecture, and StarGAN-VC architecture.