Danil Boldyrev

Hi, this is the standard behaviour of gradio below version 4, in extended I have removed this behaviour as I modified gradio. In the built-in extension I don't have the...

@tin2tin The problem is due to deepspeed, the authors don't support python above 3.9 for windows I was able to use this guide to build a wheel for windows and...

Cool, It will be interesting to look

Nice, thanks :)

Before the official comes out, you can use my checkpoint. #27

Just put it in the root folder and use `voyager.inference("YOUR TASK")` instead of `voyager.learn() ` For example `voyager.inference("Chop wood by hand, craft a workbench and create a wooden axe") `...

These tasks are embedded in the training, they serve as an example for others to perform. any task can be written in `voyager.inference()`

@smallsudarshan Hi, thank you for the code, I put everything in one place and made it easier for someone who will want to do a DVAE finetune, https://github.com/daswer123/xtts-finetune-tests/tree/main/dvae-finetune

@tuanh123789 Wow, thanks, it turns out we have the ability to fine-tune each component for XTTS. and can you tell me approximately how fine-tuning will affect the result, can we...

@tuanh123789 Yeah, I noticed that, too. Unfortunately, I haven't found a solution yet.