Neuro sama's AI singing discussions

[vast sentinel]

I just wanna make a discussion about how neuro sama has achived her singing Voice, what improvement she has been accomplished over V1 untill now and what do you guys think about behind her singing voice. is it Machine learning or voice synthesizer like SynthV, vocaloid and others.!

i made this discussion because many of us fans keeps making assumption and disinformative about her singing voice came from

maybe you guys who has gather more info with vedal statement according to this post can discuss it here!

[warm crystal]

Scientific explanation of Neuro's singing voice improvement:
#neurotic-neurons message

[lucid terrace]

Vedal isn’t willing to share this information so there will always be disinformation and speculation about this topic

[vast sentinel]

thank you i have seen that graphic as well, but i still need more info regarding how that works~ is it the same as other Ai cover or something else

[vast sentinel]

yeah i know but sometimes people push to make argument that they claim vedal did this with. which doesnt have any basis for their statement and they keep comparing neuro for using specifict Synth software which is not proven and it kind annoyed me and some make A heated comment for counter argument.

[lucid terrace]

I wouldn’t worry about it since there isn’t much we can do

[vast sentinel]

yeah and also thank you~

[lucid terrace]

Np

[nova oracle]

If you want to hear a piece of potential misinformation from me, I suspect the reason Vedal said that Evil's V3 singing voice became better than Neuro's V3 singing voice is because the project that was used for Neuro's singing voice was abadoned

[vast sentinel]

i thought evil TTS is just better to translate than OG neuro. and also both of them used the same program for their singing voices

[tough mica]

why are you guys debating, Neuros voice clone has been published for months on https://huggingface.co/yeey5/neurorvcv2/tree/main

Just use RVC to make her cover any song you want

[nova oracle]

It says v2, doesn't this imply it's Evil

[nova oracle]

Also, if it's a voice clone, does that mean it can also be used as regular TTS?

[vast sentinel]

so this is indeed an AI trainning Voice rather than Voice Synth?

how about the modified lyric is it also been done by RVC

[tough mica]

you can have it sung by someone, use vsyth, hell I can even sing as a girl although I'm a guy. This can also work as a voice changer and tts in real-time. You can also clone voices even if they are not singing, just normal speech can be turned to singing voice.

search AI voice cover RVC, there are a lot of tutorials in youtube.

[vast sentinel]

whoaa thank you for explaining and giving the Attachment, so the very based for neuro's singing is through Voice clone and for lyric modification they can rework it also through voice clone?

and for the rest of the tone it used RVC to train the Music

[tough mica]

just search youtube! i said https://www.youtube.com/watch?v=bP8AMf20MAY

[ocean copper]

Oh no. They're on to me

I've been tinkering with Neuro and RVC since June. I'll tell you it's not a simple straight forward process.

I've learned a lot

Separating out the lead vocals cleanly is a big part of it

I have found X minus is best for that . I'm also experimenting with Praat/SynthV importing to further tweak the input vocals

If yall got questions feel free to ask. I'll respond best I can

[nova oracle]

oh yeah, i have a question
can you elaborate more on how you further tweak the input vocals? do you have to deal with mispronunciations/wrong timings? if so, then how

[ocean copper]

I'm using a set Io praat based import scripts for Stnth V atm

It's similar to how vocaloid works

@nova oracle importing into Sythn V or a similar program allows you to tweak pitch and the words being spoken

[nova oracle]

i see. another question, do you also use google colab/gradio like in that video or do you use something else

[nova oracle]

also, maybe it's too broad of a question, but how long does making one song take on average, including the synthv tinkering

[ocean copper]

I"m using a fork of the Gradio project for the voice cloning yeah.it mostly depends on moviation and how hard it is to clean up the vocals into something usable

[nova oracle]

what's like the best case scenario if i may know

[ocean copper]

assuming a good extraction and a clean timber match? 2 hours

@nova oracleim actully streaming me working on soemthing ATM if you wanna watch

[nova oracle]

sorry i missed it, was busy, thanks for the info though. do you mind if i come back to ask things later just in case?