Grad-SVC

Grad-SVC copied to clipboard

如题，想请教一下Speaker Encoder是怎么训练出来的，有参考的代码吗

zyy-fc

电音现象问题请教

1

想请教一下，在经过扩散模型之前的声学模型，也就是从hubert 到 mel的这个阶段，这个出来的mel直接送到声码器，为啥会有电音现象呀，按理来说，hubert已经包含足够多的信息了，为什么生成的mel谱还有那么多平行的共振峰呢？楼主有没有试过用wavLM替代hubert呀？

postmelone

why skip_diff_train before fast_epochs

7

Hi, Can you explain why skip diffusion train before the configured fast_epochs? And how many epochs does diffusion train need? Thanks!

wblgers

训练数据量？

1

想问一下，预训练模型用了多少数据量训出来的。

postmelone

Does SVS work in english lyrics?

2

Does SVS work in english lyrics?

AbdullahMustafa040

What is the advantage for Grad-SVC, compare to So-VITS-SVC?

3

Hi @MaxMax2016 Thank you for this wonderful project What is the advantage for Grad-SVC, compare to So-VITS-SVC?

MuruganR96

A better alternative to Grad-TTS

1

Thanks for noticing Better Diffusion Modeling Technology. Recently, Xue et al. proposed that [Multi-GradSpeech](https://arxiv.org/abs/2308.10428) using Consistent Diffusion Model as the generative network outperforms Grad-TTS in both single- and multi-speaker scenarios,...

WelkinYang

What is the best version?

12

There are variety of version in Grad SVC (V3 CFM, V3 CFM RoPE, V2 96, etc..), but what is the best version?

bukhalmae145

### Thank you for the amazing open-source project. However, I am facing an issue when inference, so I have two questions. 1. Is it necessary to separate the vocals and...

ducchung2444