🧬│ai-chat

[wheat burrow]

I just downloaded all 5gb of rvc on mobile data yum

my entire phone is freezing unzipping this file

[covert lake]

which iPhone u got?

[forest quarry]

that's considering they got anything newer than the 14

the 13 has 4

i think the pro has 6

[patent crypt]

Hello

anyone knows for an IA to remember things?

[wheat burrow]

12

[forest quarry]

oh dear

[wheat burrow]

yeah

ram 🤑

[covert lake]

what

[covert lake]

I'm genuinely shocked if an iPhone 12 can't run RVC

even a samsung from 2019 could

[covert lake]

4gb of ram moment

can't believe a 2019 samsung is more powerful than an iphone 12 💀

[forest quarry]

it likely can

i just think it's because they're going out of the ordinary use... shit wasn't made for that

[wheat burrow]

I just fucked up the folder permissions

[covert lake]

6gb ram is the bare minimum to run RVC

[forest quarry]

[wheat burrow]

meh I’ll just make myself root

[covert lake]

correction, Galaxy A9 2018

[forest quarry]

i swear

I've seen a guy

do it

on a 4gb shitbox

[covert lake]

and that has 6gb of ram, still more than iPhone 12

[forest quarry]

It took a long time

But it did it

[covert lake]

are u really sure?

last time I seen someone run RVC inference on 4gb ram cpu laptop, it went out of memory

[forest quarry]

well not 100% but

I'm fairly sure yes

[covert lake]

average apple user when doing anything that isn't made by apple, lightweight or design be like

[covert lake]

do u got some proof about that?

[forest quarry]

that's problem

this was from some kid I haven't gotten dms from since like last year

[covert lake]

#🧬│ai-chat message

here's proof of it running out on memory on an i3 11th gen 4gb ram laptop

[forest quarry]

this guy tried on mangio

and i don't know the result

but it did work initially when they pressed convert

they said it was slow

that's all i know

[covert lake]

prob didn't end well

[forest quarry]

so it's likely possible yes

[forest quarry]

well i guess we'll never know

.fm

[indigo pike]

Now playing - [not] MJ

Smooth Criminal (Live)

Michael Jackson • The Indispensable Collection

7525 artist scrobbles · 2 album scrobbles · 32868 total scrobbles

[forest quarry]

oops this is AI CHAT

we need smooth criminal but ai...

[wheat burrow]

pip was going insane because I was using zsh instead of bash

oopsie

[covert lake]

don't u got a motog34 #✦│chat message

u got 4gb of ram, maybe u could try

[forest quarry]

hell nah

this thing won't make it

[covert lake]

I actually was searching for someone testing it on 4gb but couldn't find any sucessfull ones

[covert lake]

ye 😭

@night lake btw I think it's better to update the mnimum RAM for RVC inference to 6gb of ram, it isn't much stable but its doable

if RVC work on a 6gb ram phone, surely does on a 6gb ram laptop

also W10 isn't really a requirement, u can install Mainline & Applio on Linux and Mac as well

https://docs.applio.org/applio/getting-started got both linux and mac install method for applio

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md got linux & mac install method, also intel arc gpu

[edgy bloom]

Congratulations Nick088 [ITA/ENG] by Weights!

Your Charizard is now level 75!

[wheat burrow]

don’t you love pip modules not wanting to exist on arm64

[polar flax]

skill issue

[wheat burrow]

pip is 100% fucked up on iPhone

works fine on my M3 mba

[polar flax]

mac users can use most python applications wdym

[wheat burrow]

too bad I’m not doing it on a Mac and I’m doing it on my iPhone

I’ll do this when I’m home and not on the highway

[gray rover]

Well, all this rvc on android talk and yet I wonder how you gonna manage faiss and numpy and pretty much all other packages' build for arm

[gray rover]

There, finally

There was a reason pytorch mobile existed at some point
if running linux packages was so simple on android, nobody would bother with everything else

Pretty much.. all it has from Linux is the kernel and even that is heavily modified

Tbf, I'd sooner see onnx take on rvc for android, that'd be much more doable and efficient actually

[wheat burrow]

I will get this running on iPhone one day

just not today

[gray rover]

Apple def has it easier
let's be honest

[wheat burrow]

if these packages did exist for arm64 it would be much easier for both platforms

[gray rover]

I think it's a matter of time tbf

[wheat burrow]

arm will eventually take off like crazy

just not now

Apple silicon is doing good though

rosetta is what’s keeping it up

[gray rover]

Yea, could be a matter of time but there's still quite a long road ahead

[polar flax]

arm snapdragon laptop moment

[gray rover]

Look at em aethersx2, already a decent showcase how potent arm is as long you write it properly for a given platform

[wheat burrow]

meh

[gray rover]

even ability to emu x86

[wheat burrow]

Windows and arm are a little bit weird

[polar flax]

linux on arm would be better

[wheat burrow]

I mean it’s slightly working under chromebooks but tbh chromeos Linux dev environment sucks

It’s pretty usable

[wheat burrow]

tegra x1 chips

Nintendo switch and shield tv

Really capable devices

[gray rover]

Good ol tegras in general

[wheat burrow]

true

[gray rover]

I remember how much of a show Dead trigger ( game ) did

with Tegra support for reflections n shit

Good times

[polar flax]

Mediatek-Nvidia will soon enter arm laptop market in a few years

[wheat burrow]

If we continue down this path and no weird translation layers aren’t needed and we can finally have really cheap and good performing devices

ok wait hear me out

has anyone ran rvc on a Nintendo switch yet

[brisk sinew]

hi

[polar flax]

more doable on steam deck/ROG ally

[wheat burrow]

hello

[wheat burrow]

switches can still run rvc imo

someone upgraded oled model to have 16GB ram

And pair that with overclock it should atleast work

only thing is

it’s arm still

[polar flax]

weak spec tho, even average pcs can emulate botw/totk in smooth 60+ fps

[wheat burrow]

it’s not about performance, just a quick proof of concept

Dude I haven’t slept well and my sentences aren’t making sense when I double check if I read them right

I should sleep

[glad nebula]

Didnt totk have bad performance on pc because it uses astc textures?

iirc requires more than 8gb of vram to run it without oom errors

and for being able to run it below 8gb of vram u have to use a setting that causes massive stuttering everytime an astc texture is loaded

at least that was for one emu, the other had cpu decoding which didn’t caused stuttering
But required a powerful cpu

[covert lake]

I manually built some aarch64/arm64 pip packages myself

or are you having issues with other packages? I also explained how I built it for arm64 in the same guide

[covert lake]

nintendo switch got 4gb ram, it barely runs fortnite at 30 fps on shit graphics

[wheat burrow]

I should just use those then

[wheat burrow]

I’m talking about modded oled

[covert lake]

Ye I explained everything on my guide

[covert lake]

If u mod it, prob

[cobalt grove]

sup guys

[vestal temple]

Hi

[slender estuary]

Hi.

I just got a new PC, but sadly it has an NVIDIA RTX 3050 6GB.

But I hope it can run anything other than image generators just fine, like W-okada.

Since there are websites that lets me generate unlimited images using SD or Flux online for free (Perchance and Weights respectively).

[solar torrent]

That's a pretty decent NVIDIA GPU. With 6GB of VRAM, you won't be able to train any model that much, but it should be able to inference RVC, W-Okada and Stable Diffusion like fine.

[polar flax]

you couldnt afford extra two digit bucks for 3060 12 GB?

[chilly lake]

with the new checkpointing added, surely can train something like 30min set / batch 4

[polar flax]

man that card is DOA (6 GB/96 bit lol) if you could get a second hand 3060

[twilit jacinth]

I'm having issues when I listen to the AI it works but if I put it in another other program I don't like discord

It's not coming through

[burnt dagger]

a

[drifting trellis]

Does anyone know a better website than suno ai?

[solar torrent]

A website that's known as a rival to Suno is Udio.

[drifting trellis]

Where you can extend a snippet

[solar torrent]

Never used this site, so there's nothing I can tell about Udio.

[drifting trellis]

You need to be premium on udio 😭

[radiant veldt]

For what?

Don't you get free credits?

[drifting trellis]

To upload an audio file

[radiant veldt]

There is stuff you can only do with premium

Oh I see, yeah that's one of the premium things ig

[drifting trellis]

oh

[radiant veldt]

Is it free on Suno?

[ionic pumice]

ryanz no longer mod?

lol

[dense basin]

Hey guys I am training a model the accuracy of which I am getting 32 I have tried everything from tuning hyperparameters to changing model type there are no outliers or missing values in the data I am currently doing it on a decision tree. I tried it with random trees too but the same results what should I do here's the code please help me if you can 🙂
import pandas as pd
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score

df = pd.read_csv('mldata.csv')

df['Gender'] = df['Gender'].replace("Male", 1)
df['Gender'] = df['Gender'].replace("Female", 0)
x = df[['Age', 'Gender', 'Height', 'Weight']]
y = df['Likeness']

x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.3, random_state=0)

param_grid_dt = {
    'max_depth': [None, 10, 20, 30],
    'min_samples_split': [2, 5, 10],
    'min_samples_leaf': [1, 2, 4],  
    'max_features': ['auto', 'sqrt', 'log2']  #
}

dt_model = DecisionTreeClassifier(random_state=0)

grid_dt = GridSearchCV(estimator=dt_model, param_grid=param_grid_dt, cv=5, n_jobs=-1, verbose=2)

grid_dt.fit(x_train, y_train)

dt_best_model = grid_dt.best_estimator_

dt_predicted = dt_best_model.predict(x_test)

dt_accuracy = accuracy_score(y_test, dt_predicted)

dt_accuracy

[covert lake]

That VRAM is a limit tbh

[covert lake]

Weights.gg runs it on cloud

Perhaps you should be able to run wokada and maybe some SD models locally with comfyui

[covert lake]

You already talked about this in #✨│ai-help which is the wrong channel
do NOT use old YouTube tuts
Tell your GPU in #🔍│help-w-okada

[arctic verge]

@brittle turret pedofil

he's pedofil

[solar torrent]

Really? How did you know about him?

[analog cosmos]

pretty big accusation to make without anything to back it

I can also just randomly @ someone and call them a pdf

[arctic verge]

because he did

[arctic verge]

hahahaha, of course I'm not careless either

[solar torrent]

Accusing of someone for diddlering a child with no proof to validate can get you in trouble too. Just saying.

[hallow plank]

https://google.com

[polar flax]

surely a DOA card when he could instead get a second hand 3060 💀

[covert lake]

Real

I mean he could also get a 4060 ti

Depends if he cares more about saving or durability

[dense basin]

guys can someone please help me with my ml model

[chilly lake]

I'm affraid this is not a data science/ML centered place.. this discord is mostly 15-year old AI enthusiasts catfishing each other with funny voice changer program

i'm only half joking here

[dense basin]

ahh sorry wrong place

[wind musk]

Can someone please give me a free website for making ai covers due to me not wanting to pay for Kits.ai (not also wanting to use local RVC, cause my computer is having problems)

[solar torrent]

Weights.gg

[covert lake]

Kits.AI is a scam

nvm u don't wanna do it locally

one sec

[covert lake]

You can use either:

• Weights.gg
• Ilaria RVC Zero Fastest free on cloud
• Applio (ui)

Easiest possible ever: Weights.gg

[wind musk]

Can i upload my already downloaded ai models from my harddrive to weights.gg

[polar flax]

if it's not your trained one, don't
reuploading would be considered stealing

[covert lake]

If you trained it, yes

if you didn't, ehhh that's kinda considerated stealing, but ig u could do it, use it, then delete it

[wind musk]

Ir technically Weights.gg models

[limpid tide]

Weights.gg is the 🐐

Anyone who pays for kits.ai or musicfy is dum af

[wind musk]

I Know, but where can i get my files from

on weights.gg

And in Ilaria, you have to reupload the model over and over again

[covert lake]

are you sure the models you downloaded, aren't already on weights.gg?

because u download models to put them on ilaria rvc, only when they are from weights.gg

[limpid tide]

There's a download button on each model page

Sometimes u may need to click on the three dots

[wind musk]

Let me check and see if they are on the website

Does Ilaria zero keep the models in the server cause whenever I refresh the page I had to replied the same model constantly gets a little frustrating

[ionic pumice]

i think it refreshes every 30 minutes(?)

[sullen isle]

Hello, is there a way to get a free and reasonably realistic AI voice modifier?

[covert lake]

Ilaria removes the models in IlariaRVC Zero

[covert lake]

realtime voice changer for calls?

[wind musk]

That means whenever i need to make an cover I need to reupload the same model, constantly gets a little annoying

dyt

[covert lake]

Yes ofc, it doesn't have infinite storage

else it would get filled with INFINITE models and crash because storage isn't infinite

So, either buy HuggingFace Pro to duplicate the Space on ZeroGPU and have your own models here

or do it locally if you got a good pc

or use weights.gg

[sullen isle]

yes

[covert lake]

What's ur pc gpu?

[sullen isle]

How do we know? I have a laptop

[covert lake]

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

You might have GPU 0 and GPU 1

we aren't affiliated to fakeyou.com and that uses a different technology, TTS, while RVC is STS

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[valid night]

oh can you upload it to weights then?

in request model everyone ignores me

it’s better if I do it here

[sullen isle]

GPU 0

[covert lake]

yea and what is it

[covert lake]

I don't have the time to train that model sorry

either #1159289738314919936 , #1191429836321849435 or make it urself

[covert lake]

no it's not 😭

you're using the wrong channel

[sullen isle]

intel(r) iris(r) Xe graphics

[covert lake]

ofc free requests are less likely to get accepted

[covert lake]

yea as i thought integrated graphichs, which is bad

You got 2 options:

• Run it locally (on ur pc) using the CPU mode of the wokada fork which has better performance https://rentry.co/ForkVoiceChangerGuide (but im not sure how good it's gonna be for the delay)
• Use cloud (remote good pc):

About Cloud, there are different services:

• Google Colabs (4 hours daily of free T4 gpu, easy to use, require only a google account) :
  • How to use Original W-Okada's Voice Changer Google Colab (has a Guide)
  • Modified W-Okada's Voice Changer Google Colab
• Kaggles (30 hours weekly of better GPUs, T4x2 & P100, harder to use, requires an account and a phone number)
  • Original W-Okada's Voice Changer Kaggle
  • Modified W-Okada's Voice Changer Kaggle

Wokada is the program to use RVC (Retrieval-based-Voice-Conversion, Speech To Speech) Models in realtime for calls

There's the fork (modified version), the deiteris fork which has better performance

for any issues ask in #🔍│help-w-okada

[valid night]

force them to make it for me hahah

[covert lake]

😭

[ashen cargo]

How to fix this error after start GUI for RVC
ModuleNotFoundError: No module named 'faiss'

[covert lake]

do you mean https://github.com/Tiger14n/RVC-GUI/blob/main/README.md

If so, that's OUTDATED, do NOT FOLLOW YT TUTS

Tell me ur pc gpu, and what u want to do

[ashen cargo]

I launched the model GUI for RVC from Youtube video and see this error.
I want to make a cover of a song. I have audio, I launched Colab CoverGen_NO_UI_v2, loaded the model and the song, and it write that the result is saved in this path, but i can't find it on my google drive

/content/AICoverGen-NO-UI-en/song_output/znG2g3DRbM

[covert lake]

It's not in your drive, it's in the Google Colab (cloud, remote good pc service for people who have a bad pc), you have to use the file explorer on the left of it

Please tell me the tut you're following and ur pc gpu

[ashen cargo]

this tutorial Colab CoverGen_NO_UI_v2 Usage Guide
How I can to know my pc gpu?

[dapper ginkgo]

task manager

it should give you GPU 0 or GPU 1 etc etc

[dapper ginkgo]

This is my home PC so its AMD but for AI I use NVIDIA Hardware.

[covert lake]

this tutorial Colab CoverGen_NO_UI_v2 Usage Guide
send the link

How I can to know my pc gpu?
You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

[ashen cargo]

https://docs.google.com/document/d/e/2PACX-1vThk7Qo7yCWNVbxOmahl2R8_Jgi6TFuMBUIi-PWre_HIN0lFTq-dr37Rh5iJlGgYb_vFapXMHt2W8Kp/pub

[ashen cargo]

Intel Graphics 630

[dapper ginkgo]

oh its a IGPU

[covert lake]

yea that should work, i thought it was a youtube tut

[covert lake]

try this

[covert lake]

nvm it's bad, you can't do it locally

[ashen cargo]

I launched through google environment

[ashen cargo]

Thank you, I'm I found cover. but some words were badly rewritten. are there any recommendations to improve speech intelligibility

[covert lake]

wdym what words were badly rewritten

[covert lake]

Yes, I know, I was asking because if you had a good pc you could have done it locally

[ashen cargo]

in some places the words seem to be glued together and are unintelligible

[covert lake]

show a screenshot in #✨│ai-help

[cold epoch]

yall needa make an ai dave blunts plssssss 😭😭

[covert lake]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[twilit pond]

guys im taking free model requests, if you have a dataset ready DM me im willing to train it for ya

[cold epoch]

thanks bro

[covert lake]

yw

[elder willow]

where i get voices

[covert lake]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[elder willow]

Thx and I applied

I believe so. I filled out my uni

[leaden spade]

Does anybody know what a Restrained next to a model's name mean?

[chilly lake]

re-trained?

[inner frigate]

/CREATE IMAGE

[slow vapor]

Can anyone help me find something specific?

I’ve tried so so many voice models

And they all have on issue or another

I can’t seem to find any really well trained models with a semi convincing female voice

And it’s not my system either, have a powerful pc

The issues mostly revolve around higher pitched sounds when I’m speaking totaly wig out, or it sounds like I have a speech impediment, and all my r’s sound like W’s etc

[torpid jasper]

ai

[ionic pumice]

ggrks

[errant sphinx]

Big W!

[ionic pumice]

thank you mr roy

[errant sphinx]

As always, your welcome!

[twilit peak]

What's the duration limit for premium voice training on Weights? I can't upload a 45 mins file which is 5 mins over free limit

[hollow oasis]

40 mins is the limit

[twilit peak]

But it says that premium has higher limit

[hollow oasis]

nope, you can do voice training with premium only which has a 40 minutes limit

[twilit peak]

How did I train 6 models then

[twilit peak]

And have 1 regular training left, idk why I can't upload pic here

[hollow oasis]

try at here #✦│chat

hmm what u are saying is true, weights does seem to tell u can upload longer with premium

[hollow oasis]

are you suree the ones you created before without premium also had a 40 minutes limit and not shorter then 40?

[twilit peak]

I'm sure it's 40 mins because I had to cut my file

[steel pilot]

kumar sanu model?

[hollow oasis]

hmm thats weird

[covert lake]

#1191733413657464873 with @elder willow or @hidden grotto

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[covert lake]

It depends by the model not wokada, try the suggested models in the fork wokada written guide (I'm guessing you didn't follow 1 yo yt tut right) or paid models #1191429836321849435

[analog cosmos]

no

@river adder murder him :3

[river adder]

yucky

[tepid basin]

yikes

[river adder]

thanks homie sexual

[pale patrol]

https://tenor.com/view/brother-ehh-whats-that-brother-gif-6962057458982185335

[analog cosmos]

https://tenor.com/view/happy-cat-silly-cat-happy-cat-smile-gif-11748501437476694103

[river adder]

arf ^^

[analog cosmos]

wtf is wrong with people lmao

[elder willow]

Apa taa pa te

apa taa pa te

don’t you want me like a need you baby

🤡 this shit ai i can’t use it

[ionic pumice]

???

thats it you're on the list

[solar torrent]

Why do I have to adopt you? You're not my child either.

[woeful fossil]

yo am i able to import my own voices

[elder cave]

how to download voice ?

[queen kernel]

#🔍│find-models or weights.gg

[covert lake]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/You can search rvc ai voice models at:
• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[covert lake]

what are u using?

[woeful fossil]

i figured it out

thanks tho

[covert lake]

ohg lol

[solar torrent]

To train a voice model or to inference your own audio?

[queen kernel]

He already figured it out

[covert lake]

@bright stirrup I didn't even know this was public as it was a '5 min idea that I forgot' but thx

[queen kernel]

What it is ?

[covert lake]

a prototype on a fucking meme

[queen kernel]

What 🤣 it's really crazy.

I want to try it

What a roast

[covert lake]

https://huggingface.co/spaces/Nick088/RoastCalcAI

I made this just for the memes lol

just a meme project

[queen kernel]

It roast very badly 🤣 but I like it. Good job brother 👏

[elfin hemlock]

is there any ai what can copy style art?

[covert lake]

you can train a LoRA

[slim wyvern]

русские есть?

[old fulcrum]

Guys what's the best ai cover website to make covers for free?

[analog cosmos]

https://support.google.com/youtube/answer/14328491?hl=en first time seing this

[covert lake]

what's ur pc gpu first

[pliant salmon]

Does anybody know why my res is so high whenever i use the voice changer?

It jumps straight to 10k or higher and the voice is hella choppy

[covert lake]

Show a screenshot in #🔍│help-w-okada , this is not a help channel
And also say your GPU and the link of the tutorial you followed

[zinc portal]

how to use rvc v3 fork in applio colab or other colab?

[pliant salmon]

Oh mb, Pretty new here

Cheers

[covert lake]

dw

[deft acorn]

wat up

[covert rune]

does anyone have a good ai e girl voice model

not too high piched but still e girl quality

[covert lake]

https://rentry.co/forkvoicechangerguide#voice-models-to-try-out

[pale hinge]

hola

when do i get to send msgs in pic mode

[crisp tapir]

hi yall

how do i use an ai model as a cover

[night patrol]

wym

[fallen plover]

what sample rate do you guys like to train with, i myself prefer 40k, is 32K better or is 48K better, just cant decide.

[polar flax]

depend on the audio spectrogram and the pretrain availability

[ionic flower]

Loona

[solar torrent]

48000Hz is better than 40000Hz one. If you have audio dataset that is 48000Hz, you can go for this sample rate. Otherwise, you can stay for 40000Hz.

[south siren]

sup

[twilit pond]

man where do i get voice samples to test my rvc model with

[chilly lake]

silence test

[austere condor]

hi everyone, can anyone share a female preset?

[zinc viper]

https://www.weights.gg/invite?inviteCode=p6bd9

[peak osprey]

anyone have any tips on how to not get that exaperated out of breathish noise with models

[ionic pumice]

eq

probably

[covert lake]

btw do u think it's better to use fp8 or q8

made this with flux dev fp8

it took really some seconds wow

[twilit pond]

nice

[warped dune]

I have a question: How can I change words in a song? Is there a free application or program that can do this? What methods are there?

[covert lake]

you can sing the modiifed version, separate the vocals and instrumentals of the original song, convert your voice with RVC, then put back together ur modified lyrics with the instrumentals

this is the only way to do it 100% free

else use Suno or Udio which ofc have paid plans, i think they might have an option for that

[solar torrent]

There doesn't seem to be an AI tool than can change certain words from a song to another. The artificial intelligence ain't developed that fast.

[warped dune]

I think this suggested ai cover website is great, it does what it should and it is free https://www.weights.gg/de

[covert lake]

ye weights.gg is good

not sure if it modified the lyrics tho

it uses RVC too

it's our partner

[warped dune]

Thxx for info

[solar torrent]

Song generator on Weights lets you to put a link there, but its result is kinda goofy.

[warped dune]

For ai cover is this website very nice good

[chilly lake]

there are applications that can be used to manipulate the pitch, and can modify the prononcuation

but those are very small changes without changing the duration of the song

[copper sun]

I've been waiting for 3 days to see more of an overtraining pattern in the chart and instead loss/g has now started to drop even more.
I'm curious what results this model with KLM5.0 exp1 will have when I'm using RefineGAN.

[chilly lake]

exp1 is not a good starter, if you wanna play with RefineGAN use mini instead

[copper sun]

Why is that? (Not that I fully understand it, but I'll try to understand it.)

[chilly lake]

the mini has more epochs (180) but a smaller set (20h)

exp1 is like 200h and 20 epochs

[copper sun]

So because it has fewer epochs, it's worse? But if an update comes out for exp1 that has more epochs than the mini, it'll be better right?

[chilly lake]

it is a test weights

if you wanna do something, like making your own model, mini is better

just because it has learned more about the audio cloning

[copper sun]

I'll try the mini, too. Once I see it start to overtrain with the exp1. Which I think it will happen tomorrow or the next day.

The last epoch (240) saved has the best value in "Mel Spectogram Similarity" so far, 67.78%. And this is a 35-minute dataset.
And it's not the best dataset, it would need a de-plossive module for a few words.

[elder willow]

Y'all how was the instrumental and vocal separation AI on weights.gg trained cause it's so good at separating, better than vocalremover.com

The vocals sound like the ones recorded in the studio

[solar torrent]

The Weights uses some kind of UVR5 model to extract audio. I don't know which one it used, but I think some developers from Weights might know that.

[elder willow]

has anyone used ripx before?

[solar torrent]

I've never used RipX. But I use REAPER to mix tracks.

[elder willow]

im trying to remove some harmonies so my ai wont sound like its dying

can anyone show me how to use stable diffusion to make some pictures using a face reference? i’ve spent hours trying to get it to work and i’m burnt out at this point.

[chilly lake]

use ip adapter

SDXL one are lil sucky, but SD1.5 ones are good

[solar torrent]

I think Celemony Melodyne can do this, but its melody seperation algorithms don't have that much quality. The program is also paid to use this feature.

[elder willow]

im doing the beach boys if youre wondering

[solar torrent]

Some UVR5 models can extract duet harmonies from an audio, but I'm not sure which one.

[elder willow]

uvr? im new to do this

[covert lake]

Ultimate Vocal Remover

[copper sun]

I'm not entirely sure about the harmonies, but it can usually separate lead vocals from backing vocals. MDX-B Karaoke / MelBand Karaoke model has this.

[ancient swan]

Ngl q8 might be even better

At least it gives results that are very similar to fp16

[snow sedge]

q8 is better than fp8 and needs just a bit more vram, though fp8 is a bit faster

[elder willow]

With the new update I can’t download the converted vocals on the app but only the web. Why ?

[covert lake]

true

[covert lake]

will give q8 a try

found this

[ancient swan]

the text encoder is also really important

and somehow the optimized for less vram cards version sometimes works better

[covert lake]

mmm

which do u use

i just downloaded the fp8 safetensors from the comfyui guide lol

[ancient swan]

i tried both the t5xxl and t5e4 text encoders

[ancient swan]

the models? i try all of the ones that i downloaded and whichever gives best result

[covert lake]

is it better to manually setup one instead of using the default one

all i did was get the safetensors of flux lol

[ancient swan]

i don't think you can even use them without the clips, text encoder and vae

at least in forge you need all of them

[covert lake]

how tf is this shit running

I should deffo learn more before experimenting

[ancient swan]

maybe you found some model with baked in vae, and the comfy just has it's own clips somewhere idk

[covert lake]

I think so

ye

[ancient swan]

yeah

well

less fuckery

[queen kernel]

@covert lake can you help me #✨│ai-help message

[snow sedge]

There are also t5xxl gguf versions that also help to reduce memory usage

At least I use q4_k_m and don't notice significant degradation while I can keep both flux and text encoder loaded in 16gb vram

q8 flux + q8 encoder also work, but sometimes oom when rerunning inference, q8 flux + fp16 encoder doesn't fit

[queen kernel]

cann you help me ?

[snow sedge]

Not an expert in model training so cannot really help

[queen kernel]

okay ty

[azure crow]

https://tenor.com/view/lean-codeine-promethazine-sprite-gif-5718593

[covert lake]

my bad i was trying to understand comfyui

[queen kernel]

okay. i also want to learn comfyUi but not now. have a good learning

[covert lake]

ty, goodluck for training

[queen kernel]

HI

[queen kernel]

Why ?

Ookay check your dm

[covert lake]

#1159290752195633273

[kindred kelp]

Thanks wasn't sure where

[rain spruce]

anybody have a really good egirl voice model

for a medium gpu

[covert lake]

Are you following https://rentry.co/forkvoicechangerguide ? It has suggested models

If u got Wokada from yt tuts, u got an old version

[rain spruce]

i have a versionf rom 10 months ago

[covert lake]

Bro 😭

U deffo got the old Wokada

[rain spruce]

will it make it that much worse

[covert lake]

U should uninstall that

[covert lake]

In performance yes, especially on a medium gpu

[rain spruce]

aight i think i had installed an earlier version but i didnt know how to use it

[covert lake]

Don't follow yt tuts

[covert lake]

Let's talk in #🔍│help-w-okada

[rain spruce]

alr

[viral wharf]

Hey, just wanted to share my main prompt for GPT when coding...

Coding Prompt V1.0.1
This is a Prompt: I want you to do exactly what I tell you:

I want you to: Code/Plan/Help me with my Project

Name: <Project Name>
Short description: <Description>

Language: <Programming Language>
Libraries: <Doesn't matter/Only...>
Framework/Utilities: <Doesn't matter/None/Specific Framework>
Features: <Core features or functions>
Inputs/Outputs: <What the program should take in and produce>
Complexity Level: <Beginner/Intermediate/Advanced>

Additional Notes:

<Any additional context or requirements for the project>
<Specific considerations like platform compatibility, modular design, etc.>

Feel free to use it for your coding projects

[leaden jolt]

hi guys i have a question.
is there an ai that can turn male to female?

[edgy bloom]

Congratulations Diss!

Your Charmander is now level 7!

[elder willow]

Gg

[warped pagoda]

btw is there like any way I could know how to teach a bot to write jokes in a particular style?

If I have the original jokes how do i ask it to replicate the same energy or the Same Approach to write jokes?

I Have Never Been Able To DO This

[ruby wharf]

yo @dark latch juss wanted to mention

you can use the refinegan model on the appolio no ui version

and train using the klm model btw juss wanted to mention

[dark latch]

I train locally

[ruby wharf]

have you used the new codename fork by any chance?

[dark latch]

I haven't

Only Applio is what I use now

[ruby wharf]

well if you wanted to try it out, there is one avalible and its the no ui one

if ur curious and wanted to try it out

bc i did hear it is better

its free as well

[dark latch]

RefineGAN is better but it just needs pretrains

[ruby wharf]

you dont even need anything juss got limited time

[ruby wharf]

there is / you can use the KLM one on the one im using , but im guessing you choose to train locally?

[dark latch]

I have tried KLM 5 but it seems like it needs some more work

And RefineGAN isnt fully supported yet I think

[ruby wharf]

well i just saw it and it looks updated

made to support refine gan

but im not sure

[dark latch]

Yeah I'm using the experimental Applio version which supports RefineGAN but its like not supported on weights or voice changers yet

[elder willow]

I like AI characters

[twilit peak]

Any reason volume envelope on Weights can't be set to 0?

[queen kernel]

Minimum value is 0.1

[elfin yacht]

tavşan

[dim epoch]

Thats n-sane

[humble trellis]

Has anyone managed to create a glow effect inside body, eg. glow inside abdomen. in heart area etc.

Finding it really hard to prompt this with NoobAI

[covert lake]

is there an ai that can turn male to female?
You mean voice??

like do u mean u want to be trans M to F (which doesn't exist, an AI can't do the operation if that's what u want?) or just change your voice

[covert lake]

the best (IF done well) would b e finetuning an LLM

Else you could try giving it examples with detailed prompts or using system prompts

[covert lake]

that's cool, weights.gg offers a free c.ai-like experience btw

or idk if u want to do it Locally

[covert lake]

I have never heard of that nor know what u talking about, but u could see some promt engineering guides online

[flat bone]

verify me smbd please

[covert lake]

no u gotta chat and level up

don't spam

[thin sierra]

How do i make a voice model?

[covert lake]

What's ur PC GPU

[thin sierra]

Rtx 4080 i think

[covert lake]

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

Also tell me the VRAM (dedicated GPU memory) and if it's laptop

And tell me if u got multiple GPU 0 and GPU 1

[thin sierra]

It’s a rtx 4060 and 16 gbs of ram

[covert lake]

What about VRAM?

Like dedicated GPU memory

Or did u do a typo

[thin sierra]

That im trying to figure out lemme check

I was looking at the box i got a new laptop

[covert lake]

So, it's a laptop?

Laptop GPUs are weaker

[thin sierra]

Yes i have multiple gpus

Gpu 0 and gpu 1

8.0 Gb of dedicated gpu memory
Gpu Memory 15.9 gb
Shared gpu memory 7.9 gb

[queen kernel]

GPU 1?

[thin sierra]

This was gpu 0 gpu 1 is
7.9 gb shared gpu
And gpu memory is 7.9

[queen kernel]

Do you have an AMD GPU ?

[thin sierra]

Nvidia

Rtx 4060

[queen kernel]

Same as mine

Do you want to train models

On RVC ?

[thin sierra]

Yes

[queen kernel]

You can do it. RTX 4060 is good enough for training

[thin sierra]

So how do i start

I don’t know how to train voice models

[queen kernel]

Prepare a dataset

Dataset* the voice data of your character or the person who's voice you want

[thin sierra]

Voice data like a wav of the characters voice

[queen kernel]

Yes, download it from YouTube or any source you want, remove background music, reverb, eco and noise.

[thin sierra]

I got that

[barren mauve]

idk how to do that tho :/

[covert lake]

Show me a ss

Of both

[thin sierra]

Ss?

[covert lake]

Screenshot

[thin sierra]

In dms or another channel

Because i can’t post images in this chat

[covert lake]

In #✨│ai-help

Use the proper help channel

[covert lake]

U are allowed only in help channels

[fossil hemlock]

what is the best girl voice

[covert lake]

there isn't a best one

are u using the right guide https://rentry.co/ForkVoiceChangerGuide

it has suggested models

[ruby tapir]

anyone here has a pilot model? or knows where i can find ??

[fossil hemlock]

its hard idk what to do

[covert lake]

are you following a yt tut?

bc yt tuts are old

like REALLY old

[fossil hemlock]

wdym

[covert lake]

did you get Wokada from a youtube tutorial?

[fossil hemlock]

yes

[covert lake]

because they use an older version

[covert lake]

that's bad

u shouldn't have

all yt tuts are old and use the oriignal wokada, which is old and eats your performance

[fossil hemlock]

what is the

best'

[covert lake]

give me your pc gpu in #🔍│help-w-okada

[fossil hemlock]

rtx 4070

[covert lake]

let's talk in #🔍│help-w-okada

[fossil hemlock]

is it good

[covert lake]

not here

in #🔍│help-w-okada

[forest quarry]

funny how i joined civitai like 2 days ago and now im the #25 vehicle creator in civitai

happy to see ppl actually use my models tho

yeee

[covert lake]

WHAT???

FR?

Show a screenshot LMFAO

did u upload EVERY SINGLE weights.gg model u made

u surely have a HIGH passion for cars

even when i started doing rvc & loras so being kinda addicted to the "wow i made it" feeling, i never made this MUCH models in so few time😭

[forest quarry]

not yet

eventually.

i just posted 2 models

[covert lake]

that's crazy

[forest quarry]

lmao

[covert lake]

followed u

[forest quarry]

followed vaaaack

back

i cant typ

[coral marsh]

@hidden grotto

[hidden grotto]

:wave: @coral marsh, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[covert lake]

[barren mauve]

hmm

[dapper ginkgo]

How can I use civit AI? It keeps telling me to purchase "Buzz"

[covert lake]

Either do quests to get them or buy them

if u follow or like people, u get free buzz daily

[dapper ginkgo]

Dang

Thats gay ngl

Wierd system

#1159290752195633273

[frigid condor]

why dont i offer you my ass to eat for free

[urban ether]

yo any one got a good kaos ai voice i can use

[twilit peak]

And why is that? The RVC tutorial says it can be set to 0

[queen kernel]

On Applio or mainline you can set it to 0 but idk why Weights.gg doesn't have it

[sweet sable]

i need putin voice changer

[south atlas]

mhmm guys are there any AI or model that like, when i input a song (or vocal, because i can split vocal and beat), then it output full of "meow" sound of cat, but keep the same tone, pitch, melody ?

[queen kernel]

RVC can do this. Just you need a cat model

[south atlas]

umm but some how it sound like the cat singing the songs, not like full of "meow" sound

[polar flax]

search "nyan cat" model

[south atlas]

mhmm okayy ima try again when i turn on my laptop

btw which audio source should i put in (vocal or instrumental)

also any setting ? ( i am using appolio)

[queen kernel]

Vocals.

[queen kernel]

You can adjust pitch. Else everything on default is good

[fading pilot]

Hey all. What's the best (aka quickest & best quality) cloud voice cloning method right now? Is it RVC?

[ionic pumice]

ye

peakstars mentioned

[fathom saffron]

yo guys do you know where I can find a tts service with custom voices?

I remember that I accessed a site from this discord server that worked very well but I only used it twice and I don't remember what it was called

[solar torrent]

Your GPU after W-Okada. https://www.youtube.com/shorts/QzCZKa1xvkg

[sturdy owl]

Soup chat

Is there any work on AI visualising book (novel) texts?

[thick sluice]

does anyone know where to find the u i ui a cat?

[sturdy owl]

ui i i ai ui ui i ai

[covert lake]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[covert lake]

What's ur PC gpu

[covert lake]

There are different Text To Speech (TTS) AIs:

GPT So Vits: RVC isn't as good as GPT So Vits for tts, but gpt so vits (few shot tts, which means needs just a lil training for models) can't use rvc models (and viceversa), and its only limited to: english, chinese & japanese, if you wanna check gpt so vits instead, read https://docs.ai-hub.wtf/tts/gpt-sovits/

Freemium 11labs: An easy way to do TTS is https://elevenlabs.io/, you can't use RVC model on this but its a mostly premium easy way for good quality TTS

FishSpeech: FishSpeech is a 0 shot (no explicit training needed) TTS, if you got a good pc you can use it locally else use their site

With RVC Models:

RVC is natively for Speech To Speech, but forks such as ilaria rvc mainline & applio have built in tts (using Microsoft Edge TTS to make a generated tts audio, which i suggest you to choose a tts model that is the same gender and language of the rvc model you wanna use, and then convert it with rvc)

If you wanna do tts locally with RVC Voice Models (if you got a good pc):

• You can get Applio in our docs
• While Ilaria RVC Mainline here (no guide as of right now)

If you don't got a good pc you can do tts with RVC Voice Models on cloud:

• Ilaria RVC Zero (Running on A100 GPU, free fasted rvc on cloud) and the guide

• Use Applio UI Colab (with google colab T4 free daily limit gpu)

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

[sturdy owl]

So I guess not?

[covert lake]

Like making a novel ? U can use LLMs to make text and Text2Img models to make images

I don't think there's a tool that does this automatically

[sturdy owl]

No, no I was thinking more along the lines of visualizing text from a book.

Visualizing like a video

Like an audiobook that provides audio from the book text, but for video. a videobook? idk if that makes sense

[hallow portal]

how to edit lyrics in a song using ai?

[covert lake]

You gotta sing the song, separate the vocals and instrumentals,convert ye voice with RBC then put it back together

Or maybe Suno or Udio does it easier for free

But I told u the way withrvc

[covert lake]

Like a chatbot that can see images?

[hallow portal]

I would like to change just one word in the creator's song

[sturdy owl]

No, something that creates real-time images/video following the pace (speed) of an audiobook, the source being the book itself.

[covert lake]

Yes that's what u gotta do

Even for just a word

U could cut that part of the word

[covert lake]

Ehh prob not

Especially videos, they take A LOT of computing

U could generate them then put it in ur videos, but not generate new videos realtime

[hallow portal]

can this be done on the phone?

[covert lake]

Yes it can be done locally on CPU butwill be slow, cloud is suggested

• Cloud (remote good pc, easier and faster than ur PC but it's limited):
  • Ilaria RVC Zero: fastest and simplest that you can get for free
  • Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
  • Applio Colab: max 4 hours, not granted, of GPU

Easiest possible (automatically separates vocals & instrumentals) : weights.gg
easiest cloud: Ilaria rvc zero

[glossy falcon]

will rvc v3 be available on google colab too?

[polar flax]

yes, GTA 6 is on pc now

[drifting trellis]

How to make vocal models on macbook?

[queen kernel]

Use cloud services. Colab or kaggle or weights.gg

Or you can try with "replay" it's in beta I'm not sure it will work

[drifting trellis]

Google collab has time limit

Weights.gg is bad

[hallow portal]

Aren't there any simpler programs?

[queen kernel]

So the kaggle or replay is the option

[drifting trellis]

-kaggle

-KAGGLE

[rare sorrel]

📘 Kaggle Notebooks

• Applio Notebook, by Vidal Kaggle
• Applio Notebook, by Shirou Kaggle
• Music Source Separation, by Shirou Kaggle
• UVR5 NO UI, by Eddy Kaggle
• Original W-Okada's Voice Changer, Kaggle
• Modified W-Okada's Voice Changer, Kaggle
• 🆕 UVR5 UI, by Eddy, ArisDev & Nick088 Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline on Kaggle by Cauthess

Note: Kaggle limits GPU usage to 30 hours per week.

More commands: -colabs, -hf, -local, -rvc, /help

[drifting trellis]

-replay

[heady wolf]

You must be in a voice channel to use this command.

[drifting trellis]

Where can i find replay?

[queen kernel]

Google it "try replay" but it will use your MacBook resource. Local training.

It's not a cloud computing platform

[drifting trellis]

What macbook resources?

[queen kernel]

Your Mac chip.

RAM

And Storage too

Replay is a software.

Which you can install and use on your local system without using internet or cloud services

Recommend if you have good hardware

@drifting trellis

[covert lake]

Nope

Weights.gg is the easiest to convert vocals

[covert lake]

Weights.gg uses RVC

It uses the same program, except it automatically separates vocals and instrumentals

[drifting trellis]

But it only uses like 100 epochs to make model

[covert lake]

It's not fixed, they use an Automatic Overtrain Detector to make Training Easier

It depends on how it trains + your dataset

There isn't a right amount of epochs

More epochs don't mean better

[hallow portal]

but the vocals are weak

[hallow portal]

the lyrics don't match the melody

[drifting trellis]

Ohh i always made soo much epochs on only 5 minutes datasets...

[queen kernel]

Only epochs not matter

[covert lake]

that can cause overtraining 😭

https://docs.ai-hub.wtf/rvc/resources/training/#epochs--overtraining

a model could be good on 100 epochs, 200 epochs, 500 epochs, 325 epochs...

there isn't a right amount

less don't mean worse and more don't mean better

[covert lake]

welp u gotta re-sing the modified lyrics

RVC can't do that

[ancient swan]

there is this

which in theory should copy the pitch pattern aka the melody of f0 file or smth, but idunno how or where to get or extract such pitch files that would work with that

@chilly lake do you know by any chance?

[covert lake]

didn't know about that

is in the latest applio or codename fork

[ancient swan]

it existed for ages i think

i just don't knwo what the input should be

a midi file perhaps?

and it might not be compiled correctly in applio or smth cus it gives this error when you try to input anything

perhaps it's possible to do it like this, first you input an audio you want to infer, and then you input an audio from which you extract the pitch pattern and it uses that pitch pattern on the audio you want to infer

[chilly lake]

it does not work and it never worked

[ancient swan]

sadge

would've been a cool feature

[fading spoke]

roudolf the rednose raindeer had a shiny bowl, he ate honey nut cherio

[warm peak]

Quick ques, is it possible that a person with an i7 gen 8 igpu run an ai without draining performance? Or does it depend on the operating system?

[hallow portal]

in rvc what's up with the url model?

[covert lake]

an ai
which out of the 1k+ types and models?

with an i7 gen 8 igpu
that's usually not the best talking about general ai

[covert lake]

wdym

[warm peak]

is the lightest 1k?

Also; is there a performance change between linux and windows? In general ai

[covert lake]

what? no I mean't that there are SOO MANY different types of AIs, you can't expect people to know which you're talking about

[warm peak]

like gpt 4

or just any ordinary LLM

[hallow portal]

?

[covert lake]

gpt4 is not open source

[covert lake]

if even doable, would be EXTREMELY slow locally

u could do SLM (Small Language Models) which are ofc not as smart but smaller to fit it

[covert lake]

elaborate more what u mean

[warm peak]

Thank you!

[covert lake]

I'm not 100% sure on that tbh

[warm peak]

ill google and thank you for the info, its a great help

[covert lake]

I know that Linux is usually preferred for Coding/AI, but reminder it's very different than windows

[covert lake]

U could prolly use llama 3.2 3b or 1b locally

it's only 3 billion parameters, it fit even on my phone

if u actually care about quality and speed, u should upgrade or use cloud tho

[warm peak]

I kno, but my family isnt rich

Cloud i can do, depends on the price

[hallow portal]

I would like to send a photo of what I mean here but it is not possible

[covert lake]

aistudio.google.com gives 1.5k requests daily for free of gemini flash models btw

if u know coding, u could also make ur own notebooks for running LLMs on cloud computing services like Google Colab, Kaggle & Lightning.ai

[covert lake]

show in #✨│ai-help

[warm peak]

Thank you, will keep that in mind

[covert lake]

understandable dw

[covert lake]

you're welcome

[dapper ginkgo]

Just host LLAMA

Its similar

OpenAI is down rn

[covert lake]

yes we were talking about that, check full chat

[dapper ginkgo]

Once again

[dapper ginkgo]

I Will

[warm peak]

@covert lake sorry to ping you, but should i use the transformers version, or the llama base?

i see 2 different versions on hugging face

[ancient aspen]

-colab

[rare sorrel]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• FaceFusion UI, by Nick088 Google Colab
• FaceFusion NO UI, by Nick088 Google Colab
• EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[river verge]

DIlly ding, dilly dong! A new RegalHyperus drum model just released!
Lose Yourself (Drum model no. 558)

[covert lake]

could you link me the 2 versions so I can see myself

[warm peak]

Nvermind, sorry for disturbing ya, it was gated

i sent the request to meta

[burnt dagger]

Hellouder

everiyone

Que tan activos son aqui?

[dark badger]

Hi

[covert lake]

oh lol

[royal saddle]

Yo what's the best software for mac for AI cloning? I'm currently using replay and just seeing if there's anything better

Also we REALLY need more vocal AIs for harsh/death metal vocal style

[hollow ether]

Hola

[willow grotto]

Is there ai image to video that is free?

[covert lake]

text/image to video AIs:

• Locally (runs on ur pc):
  • pyramid flow (Image/Text to Video)
  • cogvideox 1.5 5b: Image to Video, Text to Video
• Cloud (remote good pc, running on an online website for example, easier to setup):
  • Weights.gg (paid only)
  • pyramid flow (Image/Text to Video) (HuggingFace Space)
  • OpenAI Sora (paid only, in some countries)
  • lumalabs
  • Hailoua AI

[willow grotto]

I love you man

[gray stone]

can someone help me, add or dm me pls

i have the ai model but idk what to do now and i cant figure it out

[covert lake]

!howtoask

[pine acorn]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[covert lake]

Elaborate on what u wanna do + what's ur PC gpu

[solar torrent]

Please be specific on which voice model you've downloaded.

Otherwise, we can't identify the issue you're encountering.

[ruby vessel]

How do I search and not get Weights links? they suck

[solar torrent]

Refresh the page and try search on Weights again.

[ruby vessel]

It only allows up to 50mb

I wanna use Applio

Wait did Applio remove its search function? wtf

[ruby vessel]

Why would I use Weights Over Applio

[solar torrent]

No, but like your question confused me when you said you wanna use "Applio", and I sent you a screenshot of searching on Weights.

[ruby vessel]

No i mean like, the search bot in this server used to not show weights links, it used to show applio links which allowed you to download the files

[solar torrent]

What?

[ruby vessel]

What are you confused about

Oh you joined October 12th, maybe you weren't there when the applio boit was?

[solar torrent]

But like how the Weights bot used to show "Applio" link to download the voice model? What are you talking about?

[ruby vessel]

I wasn't aware the Applio bot wasn't here until now

[solar torrent]

I already know Weights was already here in AI Hub by Weights.

The Weights bot here use for searching a voice model. Unless you mean the -rvc command, which Botmaze will tell you where to download Applio.

[ionic pumice]

peakstars mentioned

[soft fog]

does anyone know any ai song finder some somehting

[quick token]

is the NVIDIAs new jetson nano computer for AI find any usage? I can't think of any application that such a small computer is needed. It'll also need a lot of memory and storage for the model to run.

[polar flax]

if you can run rvc on nintendo switch, you should be able to do so

[solar torrent]

Hi.

[glossy falcon]

Will RVC V3 be available in Google Colaboratory?

[torn crow]

Hey Guys! Can anyone do the following:

1. Train a Mistrial model about Law
2. Find out an open source model like Suno or similar to that

[polar flax]

but can it run GTA 6?

[solar torrent]

The official RVCv3 won't be available as of now. Only some unofficial RVC forks that claimed to work better as RVCv3.

[polar flax]

yea half life 3 runs great on a 4090 with full path tracing

[solar torrent]

More like RTX 5090.

[hallow portal]

what to choose when I need my model to sing my lyrics (I use Weights.gg)

[covert lake]

They don't suck 😭

[covert lake]

Because Copyright got their ass, either use weights.gg or #1175430844685484042

[covert lake]

Applio bot cant do that anymore

What's so wrong with weights.gg?

It's our partner bro

[covert lake]

Who ran it on the switch

[covert lake]

RVC V3 will 99.999% NEVER come out officially, the creators almost left RVC and it's stuck at v2 since almost 2 years, however there's some experimental forks

[polar flax]

I forgor someone who tried to run on that jailbroken device

[covert lake]

Did it work

I remember someone wanted to try but didn't say after

[covert lake]

https://www.youtube.com/watch?v=7r2iBhxDhEM

[ancient swan]

ngl i feel like the dude he was talking to is just some random guy

who actually doesn't have any info

(also funnily enough one of the people who i know on a different server at some point had the exact same panda pfp, but he says that it's not him lmao)

[covert lake]

the panda pfp is to censor iirc

[ancient swan]

ah

[covert lake]

I hope so

[chilly lake]

australian prices are also weird because of taxes

[covert lake]

True

[hallow portal]

@covert lake what to choose when I need my model to sing my lyrics (I use Weights.gg)

[whole onyx]

so is this the real dc server ?

i dont wanna download anything if its not

[solar torrent]

Real DC server? This is AI Hub by Weights.

[whole onyx]

so its the real one

[solar torrent]

No, but like what does the DC stand for?

This server here is one of the only real AI Hub servers available.

[whole onyx]

so none of the ai voices will give me a virus

[solar torrent]

If one antivirus program detected a voice model zip you download from here, it's a false alarm, and it's safe. Unless you download a voice model from an unauthorized source, this can give you a virus.

[whole onyx]

so the mai sakurajima is that one safe

can u talk to me in one of the ai voices bc i am scared to download

[solar torrent]

Like what I said, these all safe if you download from this server. Unless you click a random link a random user sent to you.

[covert lake]

Yes

[covert lake]

DisCord

[solar torrent]

[covert lake]

Btw YouTube tuts are old

[whole onyx]

oh ok

[quick token]

I want to build a serverside llm or cnn application on it and host a client side on my android device. I am trying to figure out the whole chain of things, learning otw.

[whole onyx]

so if i download from this servers bot its safe

[solar torrent]

Like what I said.

[whole onyx]

k thxs i hope it is

no virus for me again pls

[polar flax]

I'm not sure but ig running 2-8b llm should be fine, so good luck

[covert lake]

Wdym again

What did u download last time

[dapper ginkgo]

Prob from a unkown source

[covert lake]

😭

[covert lake]

What's ur PC GPU and what u downloading now ?

[whole onyx]

i am downloading mai sakurajima voice thing

my pc can run very high games so it should be fine

[covert lake]

Which?

Send me the link of the tutorial

Idek what's mainsularjima, but I'm guessing a voice model

[whole onyx]

its on this dc

[covert lake]

AI usually takes more power than games,
What's ur GPU?

[covert lake]

Yes, along with other 6 programs atleast

U gotta be specific bc there's a lot of different AI programs and versions

It's not like a steam game

[whole onyx]

Nvidia RTX 4090 thats my gpu

[covert lake]

Hell yeah don't worry then

[whole onyx]

why does that matter anyways

[covert lake]

To check if you're using the wokada or wokada deiteris fork

The deiteis fork >>> normal wokada

In terms of more updatesd + performance

Also the GPU matters a lot, but yours is good dw, just asking bc usually people have bad GPUs

[worn geyser]

@elder willow

[solar torrent]

https://tenor.com/view/ping-fail-epic-ping-fail-reimu-fumo-gif-21959966

[hallow portal]

what to choose when I need my model to sing my lyrics ?(I use Weights.gg)

[worn geyser]

https://tenor.com/view/timati-angry-zloy-ne-shuti-dont-fuck-with-me-gif-11439080

да сука

[worn geyser]

пиздец

[mental comet]

У меня пк грузит. че делать?

[solar torrent]

You can use any voice model there at Weights.gg, as there doesn't seem to be any recommended model there.

[worn geyser]

плакать

[solar torrent]

Please speak English in chat here or speak Russian at #🌏│русский.

[worn geyser]

а у меня там неизвестно написано

[covert lake]

wdym exactly ?

[covert lake]

Speak english, it's the server rules

[hallow portal]

What is wdym?

[covert lake]

"What do you mean"

[mental comet]

My PC is loading, and there are a lot of voice artifacts

[elder willow]

Can someone help

i don’t know how to download this

Is it even downloadable?

[hallow portal]

so I already have the model and I would like it to sing a song

[elder willow]

Hey can u help me how do i download this?

The voice changer

[covert lake]

Sing a song with modified lyrics or just AI Cover

[solar torrent]

-realtime

[rare sorrel]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[covert lake]

sure, im guessing realtime voice changer for calls, tell me ur pc gpu in #🔍│help-w-okada

[covert lake]

they could have a decade old pc

don't just give the links

[solar torrent]

If you have a GPU, click on the first link.

[covert lake]

always ask for their gpu

[solar torrent]

No, I'm just saying.

[covert lake]

bc some people don't even know what's a gpu lol

[covert lake]

u need help? elaborate more

[modern birch]

How can I train instrument voice model for example guitar piano or violin. What dataset that I should use to train this.

[elder willow]

Im at the github page how do i download it

[hallow portal]

but I would like the voice to sing the original song with a changed expression

[solar torrent]

GPU stands for Graphics Processing Unit. A GPU processes and output image to your PC monitor, if you're wondering. There are both integrated and dedicated GPUs. However, some newer GPUs can process AI too.

[elder willow]

ight bro the video was fake asl

[solar torrent]

You wanted to train an instrumental model like a voice model? There are some custom pretrained voice models available somewhere.

[elder willow]

this voice changer here doesn’t work you can’t even download it

[solar torrent]

Tutorial videos on YouTube are outdated.

[elder willow]

How the fuck do i download this

[covert lake]

you can't, either make firstly a song you want in weights.gg but prob won't be with the same instrumentals,

or sing yourself the vocals and then convert them

[covert lake]

#🔍│help-w-okada

[modern birch]

Yes,For example if I have melody I use model to change it to piano sound or guitar. I found someone train this before I want to make this but I don’t know how to.

[normal hull]

is it normal that weights is barely loading

[urban ether]

ya happens

anyone want to here my new ai song

[normal hull]

i left it open it has been almost 2 hours 💔

[urban ether]

dang

[covert lake]

wdym

[covert lake]

#1159290752195633273

[urban ether]

wont let me

[edgy bloom]

Congratulations Fake Failboat!

Your Marshtomp is now level 21!

[covert lake]

u have to send the youtube link of it, not the file, for copyright reasons

[normal hull]

uhhh it doesnt load anything

[covert lake]

could you be more specific on what program and what tutorial you're using

[urban ether]

dang i dont have a youtube channel all i have is the file

[normal hull]

i use the site version , like after i open it , it gets stuck on the main page without anything loading

[covert lake]

our discord server got deleted last time bc of that, copyright would delete our server if u don't send it via a youtube or other social platform that folow copyright laws

[urban ether]

oh okay

[covert lake]

could you link me the tutorial/program you're using? there's tons of different ones

[normal hull]

well i didnt train or do anything to follow a tutorial and iv got no programs i use the site version of weights ''weights.gg'' on google chrome ofc

[covert lake]

ohh, so it just doesn't open the site when u go to weights.gg?

where are u from? last time idk why it didn't work in south korea

[normal hull]

it opens but uhhhhh i wanna show a ss cuz i dont really know how do i explain it more clear

[covert lake]

u have perms only on help channels, maybe try in #1192011222023950368 or #✨│ai-help

[normal hull]

alr

[quick token]

Can't be done without acquiring a lot of data of different instrumentals of the same song.

[solar torrent]

For training a voice model, you can ask someone there at #📑│making-models. I think someone there can tell you where to get these pretrained voice models to train an instrumental model.

[solar torrent]

For RVC, you indeed use a lot of dataset audios to train one pretrain model from scratch, so any voice model can be trained on that pretrained model but with less dataset for general use.

[polar flax]

training from scratch could never be good, so using any existing pretrain is the best bet

[quick token]

How can you tell. Is there a formula to determine that! Any source I can read?

[polar flax]

nah I suck at math, there should some info you can search

[hallow portal]

what should I use when I want to keep the original song vocals and audio but changing the lyrics

[covert lake]

I think I already replied to you that I'm 99% sure that the only way is u sing

Then convert with RVC

[thin ermine]

h

[covert lake]

I even searched online and even with Suno or Audio u can't bc it would still be kinda different

[hallow portal]

but singing will not play instruments

[slow vapor]

guys did i just get scammed for a voice

modelhow does this work

i paid 15$ for a voice model on kofi

never received a download link or anything

From litsa the dancer

[night lake]

how long has it been?

and proof?

[slow vapor]

Nvm we good, they did not have the download attached. They sent me a message asking if I wished for them to send me a download link. Phew

If they don’t send a link tho I’m in trouble

[ancient swan]

lmao

dw he won't scam you

[slow vapor]

Oooof they don’t appear to be responding now

He? Thought it was girl XD

[ancient swan]

just wait a lil

don't panic xd

[slow vapor]

I assume then you know the seller?

[ancient swan]

yep

a friend of mine

[slow vapor]

Thank goodness, I was so worried

[ancient swan]

and a mod here lol, so he'll def won't scam xd

[slow vapor]

His voice models are of a quality I rarely hear

[ancient swan]

yeah, he's a knowledgeable guy

[slow vapor]

I’ve been surfing through so many

So many just sound bland, have too much noise, they break on higher or lower pitches etc

[ancient swan]

yeah, making a good sounding model isn't easy

[slow vapor]

I’m a audio engineer so my standards are likely much higher than most peoples

I hear things most don’t

[ancient swan]

i see

[slow vapor]

Yeah that’s 100% understandable tho

I know it takes a lot of effort to train a really well done voice model

[ancient swan]

yeah, mostly it requires good knowledge of software like izotope RX11 and different ais that can remove noise, music, etc.

[slow vapor]

I use izotope plugins

In my daw

[ancient swan]

and good ears obviously