stop giving me ideas, I have other things to do (
)
https://www.youtube.com/watch?v=Bag5CYnXnz4
Original song: https://www.youtube.com/watch?v=YlUKcNNmywk
#vtuber #neurosama #aicover #redhotchilipeppers
1 messages · Page 1 of 1 (latest)
stop giving me ideas, I have other things to do (
)
https://www.youtube.com/watch?v=Bag5CYnXnz4
Original song: https://www.youtube.com/watch?v=YlUKcNNmywk
#vtuber #neurosama #aicover #redhotchilipeppers
What AI model did you use? I feel like the RVC model I have could do something that sounds more like Neuro
I wanted to try it because it sounded much better with a sample and added just 19 hours ago
I don't think I've seen that website before. Do you have the separated vocal for the song? I could try to run it trough the model I have
it's too big, even though it's WAV
I actually thought vaw is smaller than mp3 for a moment
Well, it's not. Wav is uncompressed
That thing sounds kinda strange, but I guess we'll see what I can get. Having the instrumental as well would be useful
The base vocal might not be that good for Neuro's voice, you could have better luck with a vocal made using some kind of synthesis program and then conversion with RVC
This is just too accented
I just wanted to do something simple, because I'm actually working on other project
Well, I guess that is just what you get with simple projects. If you want to know what the stuff I have is capable of, here's some samples of the Evil Neuro RVC I have
And here is what the Neuro RVC I have can do with your base
It's just mindblowing. I would definitely like to reach this level, but I don’t know if I’m ready to deal with it, because I'm actually lazy and don't know anything about music
You just need an application like SynthV and then somehow need to use the application to make a base, then use Applio to run the RVC model (note that the bases for the Evil RVC stuff were not made by me but made by wispers, who also trained the RVC models on data I provided, so I have no idea how to use SynthV (although I probably should learn at some point))
It's not really that complex
The main thing is to obtain a good base to work with
If you have a good base for the RVC model to infer from, you can get results of similar quality to what I have if you use the same models
The models I have are available on my FTP server
Applio is on GitHub https://github.com/IAHispano/Applio
And once you have good models and a good software to run them, you just need a good clean base
You know what's funny? I came across SynthV just yesterday, and was pretty exited, because it's actually quite easy if I have midi's
If you know how to use SynthV and Applio and have models like mine, then you can generate the kind of quality I have
If you want, I can explain the Applio and models side in DMs
So we don't clog up this thread with too much technical stuff
I would be glad, but not today, I actually should have gone to bed a long time ago
And thank you
Ok then, feel free to DM me any time and I can explain the stuff and give you instructions for how to obtain and run the models
Video is not available
It contains content from a WMG partner. It has been blocked in your country due to copyright infringement. 