最高峰のRVCv2ロリボモデル「みあ」
-
-
-
-
-
Does anyone have RVC? Do you know where I can download it to use this voice in real-time?
@le0nidaz https://note.com/light_clover642/n/ne36de5abcb7b
The installation instructions are available here, but they're written in Japanese, so you'll need to translate them.
Don't worry, RVC itself is completely free to install!
I'm Japanese, so I'm sorry if my English isn't very good.
-
Does anyone have RVC? Do you know where I can download it to use this voice in real-time?
@le0nidaz Also, for real-time voice conversion, you'll need a fairly powerful PC. An NVIDIA RTX 4000-series GPU or higher is recommended.
If your GPU is below that level, you may experience significant latency, and the real-time conversion may not work well enough to be practical.
-
Japanese voice models won't work very well for English speech, by the way, unless English speech was in the training data used. If not, it will rely on the pre-train data a lot and won't sound good. Probably sounds great if you speak Japanese with it though! (unless you have a thick accent)
-
-
Thanks for sharing, it is pretty shocking to me to see RVC model sell this expensive. I feel like it wasn't very hard to train your own. I guess it about copyright?
@tapple
I think the reason people are willing to pay such a high price for something like this is because they don't have the time or technical skills to make it themselves.
I didn't know how to make it either, so I ended up buying it too lol.
If you know of any articles or tutorials that explain how to make something like this, I would really appreciate it if you could share them with me! -
@tapple
I think the reason people are willing to pay such a high price for something like this is because they don't have the time or technical skills to make it themselves.
I didn't know how to make it either, so I ended up buying it too lol.
If you know of any articles or tutorials that explain how to make something like this, I would really appreciate it if you could share them with me!@mamana I have train two model myself, the only thing matters is the high quality input audio file, if you spend time trim audio( cut silence out), use other tool extract voice, like UVR, MSST or online https://mvsep.com/ to remove noise and light background music( don't try with singing the audio have artifacts), and raw singing( without background music).
The key is voice diversity and clean. You want it cover high and low, and you want it clean without noise and artifacts.
Then use any tool, like RVC to start training, ideally on Nvidia card that have 10G+ VRAM, downloads from GitHub https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI have training panel in it and has been translated to lots of language, it wasn't anything hard, just standard settings and within ~200~500 iteration it will sound prettty okay, further training is recommend but if you train it too much it will became over trained and you don't want that, if you have large disk just use setting to automatic save it every 200~500 step. You can stop training and if you use same name and setting it will continue but I recommend train it in one go.
You can always ask an AI for this it pretty simple. As long as you have a good audio set, follow any tutorials will get you a not bad result.
I have a 2080Ti and my model trained 1k steps and took like one day? It was 3 years ago so things might be much faster.
️
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login