bye kits.ai, I'm away. lmao what dumb restrictions are that. no, really, I will just don't use it. I hate it, every ai is totally free, then gets popular, and then the creators see a way to make a lot of mony. #openai.com, #chatGPT4, #kits.ai, etc. that's so annoying.

I'll run rvc and see how that go's, I have an integrated graphix card so I don't think it will work but let's try.

RVC gives worse results. I trained my model on voicemy.ai and then tried to make inference on voicemy.ai. kits.ai and on my local machine with RVC. Kits.ai and voicemy.ai give the same result, probably different in some small details while RVC works much worse and none of the additional parameters helped me to improve the result. However, there are some different builds of RVC around the internet. All of them are unstable in performance and difference of final results with the same models, so there is a lot of space for experementation.

Yeah I'll have to train voices on voice my and then use them in rvc

Did anyone managed to train voices using RVC itself? I have a GPU, not the newest one but definitely worth to try training on it. However, RVC behaves quite strange. It doesn't give any errors out and even kinda simulates training proces for some time, definitely very smal time for proper training though. So at the end i even got all the temporary files but not the final model.

some things I did with tts

wow, the eloquence model sounds really great though.

I like the Alex one, smile.

voice my is down right?

@144 who even is that, it sounds good though

No, I don't think so
-- (Monkey999):
voice my is down right?

--

it says on the website voicemy.ai. Is that their site?

@170 that's Garo. The voice moddel. The song I guess you know?

Also my ver sounds really similar.

Also are there any other good programs that can do similar stuff to kits?

there is rvc. But I have no idea how it works.

And I don't think this can do youtube links, or can it? Never used it though... same

ya
-- (neptune):
@170 that's Garo. The voice moddel. The song I guess you know?

--

RVC cannot handle youtube links but it has vocal extracting features onboard, so you can split instrumentals and vocals from audio files manually and then use vocals for inferring. After inferring is done you can use any audio editor to mix vocals with the music. However there are some things you should keep in mind before using RVC.
1. It is very unstable Some things which are supposed to work out from the box may not work on your particular machine because, well, because you will have to figure it out why yourself in each case. Sometimes you will just need to restart RVC or your machine because all memory handling things are written very bad. Sometimes it will be something harder to guess.
2. It is not as straight forward in its parameters as commercial services are. It is the good and the bad news at the same time. On the one hand, you have more possibilities to experiment, on the other you will have to spend more time to optain result as good as on kits.ai.
3. You need to have a reasonably good GPU. You can infer on CPU but it is very slow. As for training, it is possible only in theory but in practical use it is on the low edge of possible.
As for me, i have managed to solve most of the problems i was writing about some days ago. It didn't become more stable but i at least can do everything declared in the readme and interface.If i have some time, i will try to describe my experience and give some tips but for now i don't promise anything.