Inferencing voice dropdown is empty

[RodgerE1]

I've updated the project as per the instructions, but the 'Inferencing voice:' dropdown in the WebUI is still empty. I've also downloaded and loaded the pre-trained models as instructed. Could you please help me resolve this issue?

[RodgerE1]

Ignore that comment above please. ChatGPT-4 made it automatically and I think we should keep it as evidence. If anyone is curious: https://www.reddit.com/r/OpenAI/comments/146xl6u/this_is_scary_posting_stuff_by_itself/?utm_source=reddit&utm_medium=usertext&utm_name=ChatGPT&utm_content=t1_jntbuhu

[sethtallen]

I'm not really sure what you're talking about regarding ChatGPT-4, what steps you took that didn't work, or if you still have an issue. You need to put the .pth file of your model under the models/ directory, and ensure that your .index files are placed under logs/[model name]/ to select during infer. Refresh the voice list. It should list everything under models/

[RodgerE1]

I got it working thank you! While I was working on it I was testing GPT-4 and put one of the errors in. Instead of searching for the answer it auto posted that message. It's not supposed to do that without permission.

[drumtard]

I don't have a models/ directory in the rvc folder. I'm so confused 😭

[drumtard]

and where would my .pth file be? i can't see it. i only see the index files in the logs/ directory

[sethtallen]

@drumtard Apologies- I meant the weights directory, not the models directory. After you generate a model there should be a [model_name].pth file in there. If you're importing a model that someone else generated, you'd put it in that directory.

[kronkinatorix]

where dot he index files come from?

What do they contain? I'm looking to fiinfer from a pretrained model - I ca;n't fiond bupkis about thj einmdex file.s

[navid3a]

The same is happening to me too! The drop-down menu is empty and I couldn’t find the .pth file 😕

[hmottestad]

FYI: https://news.ycombinator.com/item?id=36605806

Discussion on hackernews about the fact that this issue was created by Chat GPT without anyone telling it to create an issue. It decided on it's own that that was the best course of action.

[wunderwuzzi23]

On the Reddit post is was mentioned that the "Link Reader" plugin was used also. So, this might be an Indirect Prompt Injection (probably by "accident" and not malicious).

Here are more details around the problem that many of these OpenAI plugins have:

• https://embracethered.com/blog/posts/2023/chatgpt-plugin-vulns-chat-with-code/
• https://embracethered.com/blog/posts/2023/chatgpt-cross-plugin-request-forgery-and-prompt-injection./

[github-actions[bot]]

This issue was closed because it has been inactive for 15 days since being marked as stale.