Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 

Repository files navigation

Discrete-Time-Diffusion-like-Models-for-Speech-Synthesis

This is the code for Paper: Discrete-Time Diffusion-Like Models for Speech Synthesis.
This repository is based on original Grad-TTS implementation by Huawei Noah’s Ark Lab. Compared with the original code, main changes lies in model/diffusion.py
Hydra is used for managing Hyperparameters.

usage

inference

Inference for four different systems:Gaussian Additive Noise, Gaussian Multiplicative Noise, Blurring Noise, Blurring+Gaussian Additive Noise can be done by 1) renaming diffusion_GAM.py, diffusion_GMM.py, diffusion_blur.py, diffusion_warm.py to diffusion.py in 'Grad-TTS_Hydra/model' folder 2) choose correponding checkpoint repository name in 'eval.checkpoint_dir' field in Grad-TTS_Hydra/config/config_eval.yaml file 3) run ./eval_all.sh

training

Training can be done by both replacing the corresponding diffusion_*.py file and changing relevant hyperparameters in 'Grad-TTS_Hydra/config/config.yaml' file.

About

No description, website, or topics provided.

Resources

Stars

3 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages