[Feature request] VITS 2 implementation

[mehrzadai]

🚀 Feature Description

VITS2: Improving Quality and Efficiency of Single-Stage Text-to-Speech with Adversarial Learning and Architecture Design

• Transformer Blocks in flow mechanism
• Text Encoder conditioned on speaker embeddings

I even love you more with these modifications ! Soon ?

[manmay-nakhashi]

@mehrzadai can you post a paper link ?

[mehrzadai]

@manmay-nakhashi https://arxiv.org/abs/2307.16430 I guess the changes will solve many things like high length dependency and some latency issues.

[erogol]

I think @p0p4k is working on it.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.

[S-Ali-Zaidi]

Hi! any updates on implementing VITS2?

[erogol]

I think @p0p4k gave up on that :(

[p0p4k]

@S-Ali-Zaidi Ill do it in this month, sorry for delay!

[erogol]

ok then reopened 👍

[S-Ali-Zaidi]

Yay, glad to hear it! Thank you both!

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.

[andersmmg]

Someone is supposedly still working on it

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.