#🧬│ai-chat
1 messages · Page 56 of 1
i still think the character design is so dumb. but i can't get her voice out of my head
https://youtu.be/WSz-eeDYua8?si=7T63QoP6P6jg0FHC and you know... its legit
Provided to YouTube by TuneCore Japan
EGOiST · Private Maniac Performer
Private Maniac Performer
℗ 2014 Private Maniac Performer
Released on: 2014-03-01
Lyricist: nakano4
Composer: nakano4
Auto-generated by YouTube.
you didn't watch the lore vid?
Didn't bother
well that explains it lol
wait is she her daughter
teto was an april fools joke on miku
hello my friends!
why is teto 30
Hello tober
you know what, let her rock it
Milf
oops
hold on
(✿☉。☉)
except she wound up being the best quality
I swear I saw that back in the day
I feel old
i first saw teto in late 2008 on 4chan when triple baka dropped
Jeez
I started listening to vocaloid last year
Aaaanyways, I watched a video and wanted to make my own model.
this song is so good
Oh god I'm getting flashbacks from highschool
@past crescent has reached level 3.
Me who discovered the existence of teto only 12 years after
the new ai vocaloids are so far ahead of the rest of the music world lol its amusing the ignorance on twitter and tik tok
Neru's so pretty
I remember someone would dress up as the matroshka miku and naruto run around
its because v2 miku and cv teto voicebanks harmonize perfectly
ya,
why doesn't Neru have a voice bank tho?
fanloid
oh
The NT miku and SynthV Teto recreation tho
Only vocal synth i've used is daisy cause its free and I'm a shitty mac user
Oh i've tried using utau and hated it
You can turn down chunks
How much?
And/or switch the f0 method
Try different values
UTAU is a pain in the ass but it can do this: https://youtu.be/qBKfeXvPaJM?si=XjX93hDFD9lJR5RC
🍒 hi there! i'm suzie. 🍒
working on this cover was a LOT of fun, thank you so much to @cyberpyunk, @Ferria, @AnthongtheGay and @Mellun-P for your help with this cover and video! it came out looking absolutely amazing!!
■Music and Lyrics: KIRA
■Art: https://twitter.com/cyberpyunk
■Video: @Ferria
■Vocal: Kasane Teto, @AnthongtheGay, @suzie-p
■M...
What that
@silent patrol has reached level 2.
Bottom right, says harvest or crepe
I make experimental music so not really looking for a synth that sounds hyper reall
Use harvest or dio
@hoary fractal
Oh ok
so I can settle for garbage lol
there is a true 100% synthesised ai voice
Oh yeah I saw it
im an ai
there's this old synth I wanna use from crystal castles but idk where to find it. It's some ancient software
use dectalk
oh that's it
https://youtu.be/nLqouKvVtH4?si=GCoMCSiAnQN1-rwk still going strong 40 years later
aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou aeiou
Check out the fanmade video that mimics the original Lagtrain's music video right here!: https://youtu.be/...
hmm its being ran through utau lol
so its a bit different
https://youtu.be/JGpf6B8EvVY?si=PTN4_bcDLlEdLz0v I want a real one
The DECtalk is a feature packed voice synthesizer and this model, the DTC01, was the first DECtalk released in 1984.
DECtalk DTC01 Emulator: https://archive.org/details/dectalk
If you want to use any of the examples from this video, I've uploaded the entire script here with some light formatting: https://gist.github.com/AkBKukU/dba8d5fb3cd428a...
Most famous synth in the world, used by Steven hawking
Is there a PC emulated version
how do i get my model maker role back?
Submit old model in #1159514067187277865
thanks!
Several versions
Awesome
no
How to use chirp command
It’s not working
It sounds like a cool feature i wanna use it
So ive done everything like the video has has shown but when I activate the Voice Changer it just echos in the Voice I have it changed to? Any ideas
Hello
what video, what software, what voice? we arent akinator ya know
@drifting summit has reached level 2.
No need for all the sas just a simple what I was talking about would have been fine
Heloo
hi
Using the real time Voice Changer Client it was echoing but now when changing audios I hear myself through the program but not through discord
@vernal moon has reached level 1.
Does anyone use a phone to train the model?
@alpine nest @hallow jewel @vernal moon
Are you serious

I'm serious
e
@lusty badger has reached level 1.
Noice
What

salken gaus
YO, quick question. I was a model maker before the shut-down how tf do get my status back without needing to go through that entire process again 🗿
No you have to go through the entire process again
im about to lose it 🧌
How do you train?
@alpine nest has reached level 5.
locally 👍
it works until you gotta restart your model over because you forgot to make sure the page wasn't inactive for too long (totally hasn't happened to me before multiple times
)
@jovial crown has reached level 1.
Never happened to me because I use an autoclicker
takes all the risk outta it, now the real is fun staying one step ahead of an automated bot and micromanaging to always stay superior over inferior machines
in my rvc beta gui i opened "EasierGUI.py" and now i get an error whenever i try to select a model? no idea what happened 
yo
@bleak mirage real
hi
wasn't there a command to get model makers role from the old server?
what software do we use to be able to use these voices?
Nope
-rvc
Hey, @untold jay!
👇 Here are some useful resources to help you learn how to make ai covers
Translation to other languages
could have sworn there was 
Upload the <t:1695670086>
Server uploaded: AI HUB (RIP)
Channel uploaded: AI HUB (RIP) Post
Profile Owner: ayymann
AyyMann
ayymann
10582
EN 🇺🇸
RVC V2
281 MB
N/A
500
0
oh dope
Try this google Colab link : https://colab.research.google.com/drive/1khsjzEigBogtkA6ZNL3gNXZqmjUAKkM1?authuser=1#scrollTo=pjlwm8SOwWVI
i enjoy clicking every link i see on the interwebs
dopenheimer
copenheimer 
copenheimer fans when seethenheimer fans show up 
hey guys! newcomer here. are some of you using google colab or kaggle to train voice models and do text to speech? if so, can you please share links for those notebooks or tutorials on how to? otherwise, i'm not so confident in my nvidia rtx 4070 - i'm afraid it will take forever to train a quality voice model or do text-to-speech.
This replacing the old ai hub server?
Ello
Also newbie here, just made my first model today by following these guides to run it locally :)
dataset creation:
https://docs.aihub.wtf/guide-to-create-a-model/dataset-creation
installing applio:
https://docs.aihub.wtf/vocal-cloning-guide/applio
using applio:
https://docs.aihub.wtf/guide-to-create-a-model/model-training-applio
thank you, about how long does it take you to train a voice model locally, and how much sample audio is that for, and on what kind of GPU? is it pretty manageable?
-colab
Google Colabs
RVC
- AI-Cover-Gen-No-UI (inference only)
- RVC v2 Disconnected (training only)
🌐 Translated colabs
- PT-BR:
-colabs_br
Audio Separation/Isolation
@torn crow has reached level 1.
😠
it depends on how many "epochs" you use, the more you use, the better your model can sound but it will also increase training time.
I've managed to create a very good sounding model with about 2-3mins of sample audio and around 500 epochs which took about 1 - 1.5 hours on my RTX 3080
still extremely new to everything though :,) so not sure if the train time can be improved
Wow, that's quite good. Can I assume that if you have ~30 minutes of audio, then you can have a lot fewer epochs, maybe 50?
@placid isle has reached level 1.
.
I'm not sure if it works that way, as far as I know, it's required to have at least around 200 epochs to have a good sounding model, I could be wrong about that though but from what I've understood having more audio doesn't cause you to require less epochs
Ohhh okay. How much longer training time would you expect if you had more sample audio? And would you expect a significantly better model if you had e.g. 30 mins instead of 2-3?
(assuming 500 epochs)
Honestly not sure since I haven't tried it yet :,)
I haven't experimented much with creating models yet, so I don't really know how much impact different audio lengths and settings have 😅
@tall notch has reached level 2.
Thanks for sharing what you know, it's really helpful for me!
no worries! feel free to ask away if you get stuck :)
Does anybody use "so-vits-svc" rather than rvc, or is rvc the standard?
pretty much every model I see is RVC so I assume thats the standard :,)
Hi
Awesome!
Nobody uses that trash anymore
guys, how to start the GUI
Quick question this is new official server right?
.
hi
hi
yes
K thx
hi
who are you
Where are you guys from?
i think so
yeah i m not so imp thats why lol nvm
👋hi
What even happened to the olding server? 500k member
holaaa
Hola
pipiska
Greetings
hllp
Do you have the output cable set in Discord
@fervent zenith has reached level 1.
What's the name of the cable program you're using
No don't
Just tell me the name
You should test if it works using Audacity
Basically you load a song in Audacity and you set the output as in input cable
If the sound goes to Discord then it works
ppl spammed illegal stuff then report botted themselves
Yeah
Don't worry it's a useful app to have anyways
What is better of these 2 Nvidia cards for training AI, a RTX 3060 TI 12 GB or RTX 4060 TI 16GB ?
@jaunty geode has reached level 1.
4060 im guessing
The one with more VRAM
You'll get faster training
If that is the most important point between both then the 4060 is the choice
Yes
thanks!
3060 ti 12 gb does not exist, only 8gb
My mistake, the model I saw with 12GB was without "Ti" 😅
i mean you know this stuff is really resource intensive
especially if it's real time
Does anybody know how to upload a png to ai-image-art?
It was first time that I saw something like this and I wanted to try that
I was unware xd
Just keep that in mind
Unless you want domestic terrorism charges on your hands (im kidding)
It shouldn't even be able to run lol
🗿
@little berry has reached level 1.
Unless you used the directML version (then it might work really badly), or attempted to run on CPU (also slow)
when I used real time it worked for about 3 secounds
After that I had to restart computer
with your setup it's best to just use the colab
I couldn't do anything bcs of lags
I tryed that one but it didn't read my input and output
colab == account ban or google changed that? 🤔
i was under the impression this doesn't use a webui
Salman
salman
bro im putting so much effort into a video no one will watch
i love starting fresh on youtube
its not done yet
it should be done sometime this year lmao!
shits hard i spent like 30 minutes on it already for 30 seconds of video
same
damn
u can later send me link on dm to ur channel
ok thank you
use dms maybe?
lol is it really worth is lmao
idk i posted it in verified you dont need to be lvl 5 to see it just lvl 5 to message there
and i freaking killed the chat there because whenever i talk in this damned server everyone runs away like i have the bubonic plaugue
that's correct, with everyone who has mental problems
but you are not polish
i was institutionalized 3 times
so it's not that bad
for an adult or for a minor cuz im 15
worst time of my life
they think its a good idea to put someone with autism in a empty room all day with nothing entertaining to do
im pretty sure i left crazier then i went in
oh btw one of the times i went to a hospital was when i threw molotovs at the police
im suprised my ass aint in jail
you see, the difference with that guy who mentioned me is that instead of being institutionalized, he writes nonsense, insults everyone and breaks the rules, then cries on this server that I'm the bad one XD
damn
i do one of those 4 things and thats spout nonsense
im a professional yapper
im pretty sure i say more coherent things when i talk in my damn sleep sometimes
polish
@little berry has reached level 2.
general chat is way more chill than verified for some reason
except for timboville hes chill despite being verified
i call him minecraft pfp guy cus thats what he is
wait my pfp is minecraft
gee whiz
poland is crazy
lookin' forward ;pp
they make interesting small arms tho
yall ever heard of the pm63 rak?
pretty interesting design
the "rak" means Ręczny Automat Komandosów
idk what that is in english because polish people dont know how the letter Z works
for some reason china made a copy called the type 82????
china steals everything so im not suprised
so basically the pm63 is a straight blowback slide opperated open bolt machine pistol submachinegun hybrid. people cant agree on what it is really
and yes i said open bolt even though it has a slide
think of how a pistol works but in reverse, the slide is back but slams forward when fired and then locks back to the rear
Yea
its in 9x18 makarov, has a rate reducer for some reason, 15 or 25 round magazines if im remembering correctly
also it tends to punch you in the eye if you put your face to close to the giant moving mass attached to the gun
@hoary fractal has reached level 18.
forgotten weapons actually shot one, a real full auto one. interesting video he also has a video explaining the history and doing disasembly
There's also theory that it was called RAK bcs the constructor Wilniewczyc died on cancer and it was called like that
Rak means Cancer
well everyone who used it said it was kind of garbage so if they ever where to use modern slang im pretty sure they'd say the thing is cancer
Yup, it was used by polish people army, Militia and ZOMO Forces
i mean obviously some people liked it but its probably because it saved a few peoples lives like any military arm would
It was called Dentist by soldiers, bcs It was often destroying teeths bcs of open bolt
yes
oh cool
i cannot think of any other gun thats both open bolt and slide opperated
maybe for good reason
It was poorly made
idk about poorly made but the design was kind of stupid. alot of good engineering is packed into it, but they decided to make it open bolt for some reason
Rak could very often reload itself
what
If u for example jump out of truck or Armoured vehicle
i think you mean accidentally discharge not reload
I'm not good and English, sorry xd
www
Its ok english is a confusing and complicated language
its like 3 languages mixed together
i think english has parts of like 20 other languages in it
If you jump, the safety go off, the bolt go back, and after that when it returned to normal position It took cartridge on its way
And could shoot you on leg
oh yea that happened on the british sten guns
something like this, I don't know how to better explain it
if you dropped it on the floor the bolt would go back enough to pick up a cartridge but not far enough to engage the sear and it would slam back forward and fire since its open bolt
Something like this happened also in RAK
and it could easily cut off your finger
when te bolt was riding when shooting, soldiers sometimes put their finger there, because of this, they sometimes managed to shoot off their fingers
@little berry has reached level 3.


Hi guys hope you have a nice day
Did some voice models get deleted?
I'm still on a quest to find silver spoon's ai model
I used to have one that i downloaded before but I can't find it at all
I just wanna do a "better than you" cover 
@elder willow has reached level 1.
There isn't one ):
OK
hi guys. Is this a original AI HUB model? SInce the old server has been struck by DMCA
I guess so
Yes
thanks
ello good people
@jovial kelp has reached level 1.
a few days ago this was a brazilian AI hub
Did they merge it or just take it over? lol
I found mine using #🔍│find-models
@robust bridge has reached level 1.
DId you try using it
man google is pissing on my parade right now
all i want to do is make tf2 mercs sing obscure japanese music and google keeps shutting the colab off
Has anyone made an ai of Shigeru Miyamoto
have anyone created a lang model for vosk speech recognition? i want to try make one for my lang, which is greek.
hi
Yo
yo
Anybody want to learn Russian?
@twin oak has reached level 1.
I only learned cyrillic
and few quotes
i mean
words
can someone gimme good girl voice?
@fluid raptor has reached level 1.
about the dead inside?
hi
not only
hi
3060 to 4090
@little berry
RTX3060 12GB are just $200-300
does anyone have an "im poppy" ai voice
does anyone have an "im poppy" voice model
does anyone have the spring voice model
I'm still a student so its gonna take me 3 months to get one
It's getting there
DirectML does the job as long as the implementation is done fine
And they are making something to port existing cuda code easily, but it's not quite there yet
At some point you should be able to run any existing pytorch code that runs on CUDA on AMD cards without any manual intervention under windows
DirectML is from Microsoft, it's just a library to run machine learning compute over DirectX12
So anything that isn't ancient is supported on windows
Yes...
is there a support chat
"but there's WSL !!1!"
@shut geyser has reached level 1.
Speaking about linux, there's a pytorch backend for AMD there, it's a tad unstable, but when it works it works
Stable diffusion works, RVC training & inference works. Just a little hassle to get working
lol
#🔍│help-w-okada for realtime voice changer
#✨│ai-help for cover making stuff
AMD's compute stack is surprisingly fast
If you ever bother trying it, feel free to ask, I spent way too much time trying to get it working that I should be able to help x)
When was that tested? I've been able to train RVC models at speeds comparable to the 3060 on my 6600xt which seemed to align with the tier
Though that was last month or so
Helo
January
I suppose it's gotten better since then ?
I don't think this was using ROCm
It was using something else to do the acceleration
Seeing how this was done on windows when the librairies weren't even released for that
@shut geyser has reached level 2.
nah, but i learnt how to read Cyrillic a while back
anyone know what the recommended software is on Linux?
@severe pilot has reached level 3.
Depends on who you ask 😅
I'd say RMVPE but there are cases where crepe does better or even harvest
hello
ы
ы y
Oh for real time ?
sup
Go for rmvpe, it usually works best for realtime, and is faster
Surprisingly no
It's just using another backend to run pitch extraction
It's mostly only useful to those running the directML build though
Basically:
- dio : runs on CPU
- harvest : runs on CPU (real slow)
- crepe : runs on pytorch (GPU for the CUDA version)
- crepe-full & crepe-tiny : run on onnx (GPU on all versions). full is slow, tiny is fast
- rmvpe : runs on pytorch (GPU for the CUDA version)
- rmvpe_onnx : runs on onnx (GPU for all versions)
U guys know how to use the models?
@modern robin has reached level 1.
I can't find the old discord
Pytorch can only accelerate on CUDA with the way the program is written
Onnx can do both DirectML or CUDA
Oh ans also, for some reason the maintainer pulled a funny and instead of disabling all CUDA-only options on the directML version, he also disabled dio and harvest
Most RVC stuff doesn't follow the best software writing practices anyways
how dare you
@peak dome check dm
Seen too much lol
a
How to let AI voice read text
I cant find mine
can someone please help me in #✨│ai-help
Hi hi hi!
is there any way i can make the voice doesnt sound robotic?
hey guys, im new to this so do you know how to do an ai cover? is there a good tutorial?
for some reason when i downloaded it, it said it was a virus
yeah ima need someone to guide me through downloading this
when i open voice ai it keeps on 0%
-rvc
Hey, @dull escarp!
👇 Here are some useful resources to help you learn how to make ai covers
Translation to other languages
why is english an american flag?
It's the better English
thats like saying canadian french is better than french
Also england created english
hey could you help me download it cuz it keeps saying its a virus
I'm a sick Gyatt I like quick rizz I like I show speed I'll buy you a Fanum tax I'll buy you some skibidi I'll get you that grimace shake how you became a sigma l've been watching Kai cenat
Bypass the warning
are you sure its not a virus ( i will do it but my antivirus will stay on just in case)
it isn't
We've been using this shit for months kid
alr mb then
im just paranoid of viruses so yeah
let pump cook
@elder willow has reached level 2.
p
Lucifer Michaelson model pls
is this your own model, or someone else's?
hellu
дароу
yo
hi
am i allowed to save the collab to make it save my music ai making?
im using someone else model
hi
by virus, do you mean "disallowed code"?
No that's a Colab thing

Can someone get me the name of yhe website that lets you download the file to make ai songs
@flint merlin has reached level 25.
hi
Any way to save a Collab without it Deleting ?
hi
Hey
I tried turning off discord eco and supression options but still the same
I like using the cutest robotic voice
Do i have to repost the models i've published in previous server? there are models i made that aren't reposted
Xd
Idk if the problem is the virtual cable or any discord config
@empty berry has reached level 1.
@arctic haven u alive?
hi!
Hi!
who cares american english is better
Amplify
English is the worst bastardized broken language
o/
how to train a model ?
guis, how to create .index file? I already made a pth file, but for some reason there is no index
where did you do your model?
It’s in the logs folder
which?
colab
America after changing a few words: 
You're litterally speaking it right now
is it your first training?
I feel like I'm the only european here
no
i have trained one successful but now it may got updated and I don't know how to use it
@mortal wasp has reached level 1.
Hello
hi
just repeat the same thing you did the first time
@knotty ravine has reached level 1.

but don't change a google disk
ok
maybe I just lost the old colab link
When you are training the same model not for the first time, then uncheck pre pth
can someone give me the link to the colab for using the voice clones? thx
i can't remember what the link was
Yep
-colab
Colab suggestions for @light path
RVC
- AI-Cover-Gen-No-UI (inference only)
- RVC v2 Disconnected (training only)
🌐 Translated colabs
- PT-BR:
-colabs_br
Audio Separation/Isolation
also is there anyway to use RVC realtime on linux? it worked fine when i was using windows but i'm using linux now so idk
That's how I know it's a busted piece of shit of a language lol
Then go move to russian or sumthing
hi
hello
wasn't there one with a webui?
@light path has reached level 1.
Not anymore
what happened?
hi
Google said NO GUADIO NO FUN NO AI STUFF
I'm studying Japanese and while there is some weird quirks it's a very structured language
so english is ur first language?
Though if your using it for covers I would use wordband instead since it’s more user friendly
Hello hello
hai
Nihon ?
Yeah. I told my coworker about "baka mitai" and he keeps shouting it to try and make fun of me. I don't have the heart to tell him they are calling themselves a dumbass. I laugh inside every time
Speaking of Baka Mitai, i made a ai cover as a joke
Kinda the point of this server right?
Yeah, just wanna slide in the convo
skill issue
This only exists because of my partner. You know who you are.
Credits:
Vocals: TETO AI
VSQx, Tuning: Haru0l (edited by me)
Mix: Me
Song: Mitsuharu Fukuyama
Video, Illust: Me
Hey that’s @wintry sinews line!
You can’t post images
Skill issue
DM I’ll see if I can recognize it
Not sure
🙁
Hey everybody I’m new to this sub, been trying to learn and explore
Quick question - as an English speaker and singer, if I use a Dutch/Japanese voice model, will it sound weird?
hey @hollow tangle , what do you think of this?
If you sing us a song , I think we will figure it out :3
ahh cant upload files
was about to upload a tts from bark-small to ask you if it would be good for my "live-dubber"
I'm pretty good at Japanese 
where do i find the link to the colab
-colab or #1159949278270193734, no webui because Google banned it
Colab suggestions for @sweet agate
RVC
- AI-Cover-Gen-No-UI (inference only)
- RVC v2 Disconnected (training only)
🌐 Translated colabs
- PT-BR:
-colabs_br
Audio Separation/Isolation
alr
Quick explanation pls
What are those link for 
Hi
Hey all good morning. Question. What is the file type for the voices that are included with the downloads. Is it .rvc ? Also how do I get to be a creator and what do I need to convert. I have a lot of fun voices to contribute but just need the means.
Models are zipped as .zip
The zip file contains a pth and index file
index usually controls accent
Usually bigger then the pth if it was a big dataset
Pth got ya. So to get a WAV to the PTH what do I need then.
RVC
I would do mangio
No limits, free, open source
Download the infer zip file
Extract using something like 7z and open the bat file

Thank you mr Minecraft brown hair
💀
nice
Can you recommend me a site or program to make an AI cover?
Mr Minecraft brown hair said mangoirvc , check that out
@tight tiger has reached level 3.
If your pc is decent use Mangio RVC
if your pc is bad use WordBand Bot in discord
hm
I found my pc in my basement 🙁
I understand ty
Is there any website for that 
idk
Is there a link where I can access the bot?
how can I use this bot
@upbeat mortar has reached level 1.
hello guys. Is there anyone who has CLA Vocals plugin cracked?
Hey thx guys u rock. Hope to be getting some stuff up here for yas to enjoy.
In case , if you failed making AI cover or somth , DM me
@obtuse wigeon selam
I will sing you a song instead
ElevenLabs I like.
@bright crow has reached level 1.
Can I DM you?
of course
Check dms
yo
Do I understand correctly that most of the models are uploaded on huggingface with openrail license.
And this means that the voice created by them can be legally used for my own purposes, e.g. for voice-overs on YouTube, right?
hello
Is that for me?
Are you sure?
You can't download the Models there..
oh, thanks my g
@exotic elk has reached level 1.
My voice Ai is not opening can anyone help?
hello, I have a question if its okay. Is it possible to find a voice model that is compatible with Sapi5? any hints or clues will be much appreciated
Hey people
WordBand Bot GUIDE BY TIMBOVILL The WordBand bot is a powerful Discord bot capable of creating covers using RVC. Logging in Wordband does not work without logging in which you can do so by using the /login command, it takes less than a minute. How to Cover It is recommended to use the bot in ...
How y'all doin
@spiral owl has reached level 1.
hello
hi
I'm sorry, of course, but where is the guide you can check some kind of otherwise I don't understand something
-rvc
Hey, @untold urchin!
👇 Here are some useful resources to help you learn how to make ai covers
Translation to other languages
can i use RVC model and use it to create a custom sapi5 voice?
can someone help me download the voice changer?
The old AI Hub get deleted.
This is the new AI Hub.
oh
hello
thanks, is there any way i can get my roles back?
You can get that back.
so do I have to post models again or no?
@mellow juniper has reached level 1.
hey thanks
Normally you shouldn't have to.
There's a bot that reuploads old Models but it's on pause...
Do /usermodels and your name in #1159514067187277865
so anyone does have any idea how do I use a pre trained model like rvc models and use it to create a custom sapi5 voice haha i dont know where to start i wanna know the format that i need so that I can use it for sapi hahaha
@meager pendant has reached level 1.
You can reupload them if you want once you get your role back.
Others are doing that.
Okay thanks
👋
👋
👋
👋
👋
👋
👋
👋
👋
@empty raptor has reached level 1.
hey gang, relatively new to AI voice stuff. What's the general best vote when it comes to RVC2 models that aren't web based?
I noticed there's a few different versions in #1159949278270193734 but I'm not sure on the difference and wanna get some opinions
can anyone help in dms how to make an AI Cover pls 😭
👋
I need some help too
Hey, sorry for being goofy ahh. The reason I joined here is just because I was looking for that Amy Rose model from app kits ai that is now private because app kits removed the community library.
Kits sucks
Highest quality methods is local rendering where you can control the parameters directly and master the audio. Or just use professional ai voicebank software (paid)
Tetski teach me how to RVC 🥺
Bro that's the only thing I could use.
The characters doesn't sounds like robots because you use an actual audio sample and convert.
Bruh
Install local from the guides if you have a powerful gpu
does anybody use local RVC2 versions? What's the difference between all the different RVC models?
@abstract pier has reached level 1.
Teach me how to local cuz how much ever i try it keep saying error 🥺
@honest apex has reached level 2.
Post specific errors with details in #✨│ai-help
Or just go pro
bro..
just teach me 😭
I'm at work on my phone. Not at my computer to do specific troubleshooting.
is there a way to see all the most popular voice models?
@distant hollow has reached level 1.
or sort them as most popular or most upvotes, rather than most recent
👋
Hey guys
based
Hey Mates
How can I make a quality cover?
guys
@fallen tangle has reached level 1.
pls
Check on #1159513888199540817
for the voice models what does epochs mean? does more epochs mean better
hi!
👋
👋
haiiiiii :D
Artificial Intelligence or “Machine Learning”
yeah but how do you do it
how long does it take to train a model
Using a fork of stable diffusion
I love hate keeping 
Depends on epochs and dataset length
epochs 300, dataset length around 9 minutes
which one, illusiondiffusion?
@flat matrix has reached level 5.
About a few hours
If your using colab
i am using applio locally
since the server's being more active, i think i'll ask my question again
What’s your GPU
what ai voice cloning software do you guys recommend for Linux?
@worldly sparrow
laptop rtx 3050ti
@worldly sparrow has reached level 1.
thank you bot, very cool
Should be faster
i dunno man, the math doesn't add up
by few hours do you mean 15 hours
there are 300 epochs, every epoch needs like 3 minutes or so
NO
And an epoch for 3 minutes should not happen for a 9 minute dataset
Few as in 3-4
Hey!
sup
hey
Ey
how's things?
Still RVC
isn't there like a specific fork?
i've just been using Mangio-RVC for the past month
does it work on Linux
yep
well it did
i've just been running it on my local system
h i g h q u a l i t y 9 minutes
Is anything else open
does the quality and size of the voice files matter
uhmm, discord, chrome, that's about it
Length yes
Close em
Clogs up ram
That’s why it’s taking so long
i saw something called applio, but in the installation process, it suggested running a .bat file, (basically the Windows equivalent of .sh, but it didn't suggest a .sh to run)
wait nvm i looked in the wrong spot
is it better if i do it on collab
like, will the quality of the model be different or
@storm meadow
It will have same quality if you use same settings
Colab would probably take longer
i will try colab
if it takes as long, then i'll just run the thing locally overnight
guys i need a durk ai voice model please
Galestrike
@idle light has reached level 21.
Yipeeee ^-^
hi guys! my Name is galestrike, big fan of super heroes!
hello
Can I get the link to the website needed for sound modeling? Mine is broken.
I'm sorry, I've tried everything 😭
It seems that I can't find any model with your search, please try another query.
I'm sorry, I've tried everything 😭
It seems that I can't find any model with your search, please try another query.
L
whats that
**Method 3 🔎: ** The search accuracy is 40% 🥺
Upload the <t:1686104223>
Server uploaded: AI HUB (RIP)
Channel uploaded: AI HUB (RIP) Post
Profile Owner: kaengpil
felix
kaengpil
217
TR 🇹🇷
RVC V2
126 MB
N/A
250
0
this is NOT swae lee
Upload the <t:1697823280>
Server uploaded: AI HUB
Channel uploaded: AI HUB Post
Profile Owner: dee_oxi
Deeoxi
dee_oxi
16776
Unknown 🇺🇳
RVC
59 MB
N/A
300
0
Hi
Like tts?
I want to make A.I voice
@elder willow has reached level 1.
#1159513888199540817 are a good starting point
That would mean you would have to train a model
Okay
#1159513888199540817 local if your pc good or Colab if your on mobile or pc bad
Can you teach me?
help me
Not right now I’m at a store
@storm meadow sorry if it's a dumb question but i still don't understand the purpose of epochs, is it better to have more of them if my dataset is about 9 minutes?
@worldly sparrow has reached level 2.
-colab training link
Colab suggestions for @fallen tangle
RVC
- AI-Cover-Gen-No-UI (inference only)
- RVC v2 Disconnected (training only)
🌐 Translated colabs
- PT-BR:
-colabs_br
Audio Separation/Isolation
thank you bot, very cool.
More does not mean better
soup store?
But can you?
Not right now
so is this like a weird balance thing i have to find on my own
Yeah, if you have a really good dataset it shouldn’t matter
thx
Just do 250 or 300 and see how it turns out
i'll do 300, thank you for your help 
To sing or talk
Talk
Elevenlabs or the Google tts, there's also a ton of programs you can install
hi i need suggestion for cute voice
Why do you recommend everything but RVC
More epochs = better (mostly)
I use those as base audio files to input to rvc if you haven't noticed
@storm meadow i need help with smth
sorry for the ping
but what pitch extractor do i use
RVMPE
why tho
Too many epochs and it overfits and sucks
I use synthV as a realtime input to rvc to control pitch and parameters directly on any voice of my choosing.
There's a few people here who do the same thing, it works very well
but what about this
Tim imma destroy you
spreading misinformation ong
Works with trash datasets
Person says the changer was slow
Honestly I've had better results with mangio crepe on most voices
bruh you just called my shit trash 💀
It's just slower
No
do you even know the difference between the two?
80% of modes use rvmpe
ok
Mangio Crepe is for studio level stuff
where'd you get that info then
I don't bother with it unless it's good quality lol
Look
thats 98.42% not 80%
ffs new people are so stupid
HEYO help me
ok Armin :3
@gaunt flower has reached level 1.
Lol what's wrong with these stickers
@compact vault has reached level 1.
Look at all the models, they mostly use rvmpe
Since it sounds the best with most datasets
So explain to me the specific differences between rmvpe and crepe because I've had varied results
Can i do it here?
i need e girl voice ffs
whos that
she looks cute
where is her model
blud doesnt know teto
Feel free to support me on Ko-fi: https://ko-fi.com/adamedmond
Uploading this first because DEATH cover got blocked many times, probably because of the music video thing so I'm doing the plan B video right now 👍
SVP distributed as well if anyone would like to try it out with other voicebank! Feel free to use it and don't forget to give credit w...
She's guaranteed to get you views
really?
Well she's been popular for 15 years now
what does that mean :3
Can sing any language
even Estonian?
Hmm I'll look
i love lurking in chat ong
weirdo
you're a weirdo
No but I'm sure it can be done, I've heard a Mongolian teto
hmm
nice
DISCLAIMER: I do NOT own the PV used for this video. The original Miku PV belongs to Anamanaguchi, and this Edited Version was done by Anh Duy. Please don't think i made this. I am taking NO credit for the PV.
Your the one asking for a e-girl model…
**Method 1 🔎: ** The search accuracy is 100% 💕
Upload the <t:1687135637>
Server uploaded: AI Hispano
Channel uploaded: AI Hispano Post
Profile Owner: troqueideaccmigoss
troquei de acc migos
troqueideaccmigoss
424
ES 🇪🇸
RVC V2
0 MB
null
100
0
I don't even care about the character. I love the voice itself.
well
so what
every epoch takes around 3 minutes, looks like i'm keeping this open all night
why does everyone want e girl voices ong
The best one is already here
is it weird i want my own voice
nope
Nope! Making your own voicebank is quite popular and encouraged
no no, i want to use it for real time convertion
Uou have a variety of methods too
voice banks are hard to make 
whats a voice bank

its a bank but for voices? 
hiiiiii
a bank
would they buy my voice :0
Think of it as a "voice library"

so an archival thing?
There's an index file like a card catalog

why my res ms so high 😦
Basically yes, organizing the audio data so it can be retrieved and applied as you like
sounds hard to make
Why real time changer your own voice to yourself
curiousity
Rvc is the easiest method but the least amount of control

its basically adding noise but sure you do you
and engine noises ofc
how can i fix this
but but

every epoch takes 3 minutes
right
Close discord and chrome
me when 4180293 hour dataset
bad suggestion but sure
i have 9 minutes of dataset though 
If you want to make an ai voice you can write your own lyrics and tune there is diffsinger. Similar technology
I AM ON MY PHONE
AND IF I CLOSE CHROME I CLOSE APPLIO
laptop rtx3050ti
That's enough
does he never do that
she*
real
check task manager @worldly sparrow
when training of course
real
A SHE?!?!?!?! well hello there beautiful
wanna hang out withbme after i train my slave... i mean my own voice model?
Ai hub is not exclusive to rvc fyi. There's. Many flavors and forks
wha
am i supposed to look for smth
ctrl+shift+esc or literally search for it
then go to the performance tab
i am on it and on performance but what am i supposed to make out of this jibber jabber
all i see is that it works 
https://youtu.be/VcGYUbeAiLg?si=844EDn5JwBAHInnX ooh this one just came on again
After some months, I'm here uwu
I love this song! it's so punchs punchs 💃🕺
✦ Credits:
■ Tuning & mixing ▶ Teto Chan!
■ Music by admk. & lyrics by 混乱P ▶ ft. GUMI
└ https://youtu.be/Dvq3zriHx0U
■ Singer ▶ Kasane Teto [ 重音テト ]
■ UST ▶ akem
└ https://www.youtube.com/c/akemiyuyu/
■ Join our Discord server
└ https://discord.g...
its a graph
the higher up
the more of that value on the y axis
simple :3
me making it more complicated

