This repository provides a multi-mode and multi-speaker expressive speech synthesis framework, including multi-attentive Tacotron, DurIAN, Non-attentive Tacotron, GST, VAE, GMVAE, and X-vectors for building prosody encoder.