yl4579/StarGANv2-VC

Based on the README, here's a technical summary: Built on StarGAN v2 architecture with a style encoder for disentangled speaker and style representation, combining adversarial source classifier and perceptual losses to enable style transfer (emotional, falsetto) alongside speaker conversion. Integrates Parallel WaveGAN vocoder for real-time inference and includes pre-trained F0 and ASR models for pitch extraction and content preservation across languages and singing conversion tasks.

518 stars. No commits in the last 6 months.