✨│ai-help

[digital cairn]

well for server solutions nvidia is great, but if you use linux as desktop OS you are bound to run into problems (or atleast more than with AMD)

[viral mason]

that dot I accidentally put annoyed me

[low shard]

yeah u prob gotta update if you're looking to do local AI tasks or demanding games, 6gb vram

[viral mason]

https://tenor.com/view/zhouzhou-orange-cat-cat-with-orange-sleepy-cat-gif-14937184825975080647

I got this chat

[low shard]

I use windows so I'm not really aware of that, tho I don't usually hear some of my friends on linux complain about issues

2k+

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[digital cairn]

depends which one you are in the market for

if you switch to virtual cable as output it logically goes into the virtual cable

then you need to record in audacity, or mic test in discord with the virtual cable output whether its working

[low shard]

that tutorial is outdated asf, it uses an over year old version of original wokada and vb audio cable which causes issues

[digital cairn]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#audio-setup

[low shard]

you can simply uninstall everything and forget it existed lol

[viral mason]

https://tenor.com/view/sick-tricks-with-wario-wario-drowns-gif-1832047849940242039

is it trolling like what I do or are you a freaK

oh they got deleted

XD

[low shard]

catfishing trying to get some free nitro should not be normalized 😭

[viral mason]

it's definetly for losers

no doubt

nitro isn't that important besides having a cool dude next to your pfp like what u have

and making the upload limit actuall exist

[pseudo swift]

😭 you could prob get a shitty 2060 for $200(?) though

[viral mason]

I want the good stuff

[pseudo swift]

ahh yep thats $$$ then

[low shard]

they want the good stuff 🔥

easy way to burn money: buying gpus

[viral mason]

https://media.discordapp.net/attachments/1024110975265292389/1114446589784752198/E59321BC-A727-4EC4-9204-4DD36F4ABEEF-1.gif

[pseudo swift]

covid and crypto fucking prices up

[low shard]

https://www.workstationshop.it/nvidia-h100-nvl-94gb-hbm3/ 40k Euros

[low shard]

real

[pseudo swift]

LOL train models in 5 mins hopefully

[viral mason]

I use real money

what's that in US $

[low shard]

around 40k iirc

[viral mason]

ah so same price

[low shard]

i mean there isn't an huge difference between euro and us dollars

im just guessing tho

[viral mason]

o

I'm a dumb american so I'd probably believe you if I didn't know

[cyan pasture]

can somone help my voice is like 20 seconds delayed and only rearly pick up my audio and changes it to a other voice

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[low shard]

original wokada is outdated, don't use it

are you trying to use a realtime voice changer for funny roleplays like "peter griffin", or for e-girl catfishing/trolling? I need to know to also help you setup realistic settings

has worse quality and performance lol

i can help you up, are you going to use it for e-girl trolling/catfishing or roleplay?

I just need to know that

alright, so no e-girl trolling in games/vc right?

great as catfishing isn't a good thing

uninstall everything, delete the wokada folder, uninstall vb audio cable from windows app settings

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

read up the wokada deiteris fork guide, there isn't any updated video tutorial for voice changers

[simple ore]

@novel summit make sure you grab 5000 series app for your 5090

[low shard]

50 series users using normal nvidia one, and normal nvidia users using the 50 serie one /j

[simple ore]

2nd one is fine

wont break

[low shard]

oh right lol

[simple ore]

probably can just recommend 5000 series for everybody

[low shard]

I mean it was supposed to be a dev build iirc

[grizzled sapphire]

can i have the download link for vac lite?

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable

[grizzled sapphire]

i dont see it

ohhh nv,

nvm

thx

is it possible to run 2 wokodas at the same time with a 1660?

[low shard]

No, why would you even?

[grizzled sapphire]

idk 😭

to like mix two voices or something

[low shard]

there is a "merge lab" button for that iirc

[grizzled sapphire]

alr thanks

[misty mantle]

Yo how can I make my own rap using another rappers voice

@low shard

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do

[misty mantle]

I wanna make a song using another rappers voice

I have the lyrics and the voice only file

[low shard]

what's your pc gpu? operative system?

[misty mantle]

I'm on mobile rn

[low shard]

damn, AI is complex and takes intensive power, you will need to use cloud computing with limited free time

you'd need to sing the rap yourself, then use rvc to convert the voice

Train (make) RVC Models on cloud:

1. Prepare the Dataset
2. Setup RVC:
   Choose a cloud way to use RVC,

• Google Colabs (max 4 hours of daily T4 16gb gpu not granted for free, not much hours for training, but easy to use, there's a paid tier):
  • Applio (UI)
  • Mainline (U, broken right now)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus, either T4x2 16gb each or P100 16gb, only free):
  • Applio (UI)
  • Mainline (UI, broken right now)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly, Free Studios run 24/7 but require restart every 4 hours. There's a paid tier):
  • Applio (UI)
  • Mainline (UI, No guide as of right now)

3. Be sure to know about the tensorboard

Google Colab = Easier but risk of getting disconnected
Kaggle = Harder but way more gpu time
If you are looking for the easiest way and for free, is using https://weights.com/ which ofc uses RVC

RVC Inference (use models) on pre-recorded audio on Cloud

You can use either:

• Weights.com: Easiest Possible Ever Automatic
• Ilaria RVC Zero: Fastest free on cloud
• Applio UI Colab: RVC Fork with some extra features like TTS
• RVC AI Cover Maker UI: Automatically Separates the Vocals and Instrumentals, converts the voice and mixes them back

[misty mantle]

😭

[low shard]

No, it's not as easy as just putting the lyrics and selecting a model lmao

[misty mantle]

I see

And ion understand

[low shard]

first of all sing the rap song yourself

[misty mantle]

K

[low shard]

then read https://docs.aihub.gg/rvc/cloud/applio-colab/

[viral mason]

it is on pc

[low shard]

We are talking about custom songs singed by rappers

[viral mason]

?

[low shard]

It's not ai covers

It's a custom song by him

that he wants to be singed by someone else

[viral mason]

like he's wanting to put an ai over it or

a real person

[low shard]

share a screenshot of your wokada

!give-media-perms 1h @raw heath

so did you delete everything you had?

[sterile yacht]

How much hardware do I need to train a model?

[shrewd jasper]

anyone know a way to use amd ROCm on windows?

[knotty moth]

wait for ROCm 7 and support by more pytorch based applications

[shrewd jasper]

ah ok :/

also

[simple ore]

depends on your GPU

[shrewd jasper]

do you know a good place to get TTS models?

[shrewd jasper]

amd 7800xt

[simple ore]

you can run a bunch of tts with zluda

[shrewd jasper]

oh thanks!

[simple ore]

for chatterbox https://github.com/resemble-ai/chatterbox/issues/52#issuecomment-2925328247

[shrewd jasper]

thanks

[barren magnet]

I need help, i don't know how to make a cover

[simple ore]

there's also a beta of windows pytorch build

https://github.com/scottt/rocm-TheRock/releases/tag/v6.5.0rc-pytorch-gfx110x

[shrewd jasper]

thank you!

and also do you know a good place to find TTS models?

[simple ore]

most tts come with their own models

it is really important to mention the language you need

chatterbox is english

there are others with more / different languages

[shrewd jasper]

are they sort of like rvc models? can you just download different voices or is it smth completely different?

[simple ore]

chatterbox clones the voice using a simple 10s sample

there's nothing to train

[shrewd jasper]

ah cool

thank you

[viral mason]

can't get anything good like rvc can from 10 seconds

[cold shoal]

Hello i have a Stan ai model (South Park) and I wanted to send it here to get anyone’s opinion if it’s good and should I use it or not?

[flint vapor]

You don't have to send it here, you need to submit the model with a sample and a staff will review and approve or decline the model and give you the role for upload

https://discord.com/channels/1159260121998827560/1305527335646269440

Here

[cold shoal]

I’m sorry my apologies

[flint vapor]

Is ok don't worry!

But the model is yours right?

[cold shoal]

I got it from weights

[flint vapor]

Oh so if is not yours you don't have to upload it

On there

The models should be trained by you if you want submit in the channel for make listen and reviewed

[peak pier]

can anyone help me with the delay?

[flint vapor]

Yes of course

Which is your gpu?

[peak pier]

RX 5600XT

[flint vapor]

Which chunks and extra did you put?

[peak pier]

it is default

[flint vapor]

140 - 200ms + 2.0s extra

[peak pier]

i just installed it

[flint vapor]

Try to use a value like this

[flint vapor]

You installed the deiteris fork or the mainline wokada?

[peak pier]

deiteris

[flint vapor]

Ok nice

[flint vapor]

You can try this 😄

[peak pier]

if i put the extra 2sec then the delay is more than before

[eternal pilot]

Staff?

[flint vapor]

Keep it 2.7

[peak pier]

thanks

now it is working pretty well

[simple ore]

????

[peak pier]

which setting should i use of mainline wokada?

[clear citrus]

can someone help me with my vc

this working?

[flint vapor]

Don't use mainline is outdated

[flint vapor]

Which tool are you using?

[craggy slate]

hello, can someone assist me with some tidbits about terraform, and governance

[flint vapor]

If can i ask you which is your purpose for using the voice changer? Is for troll some friend, play in videogames and prank people in vc or catfish with female voice ?

[flint vapor]

I'm sorry, this is not really my area and i don't know how to help you😢

[peak pier]

going to troll people in valorant also cs2

[flint vapor]

Ok ok, so not catfish with female voice?

[peak pier]

going to use omen's voice while playing omen

[flint vapor]

Ok ok Nice 😄

[craggy slate]

@flint vapor its cool - thank you, perhaps if you come across someone. im also looking for some information on ray usage with dqn, in correlation to terraform and governance usage. I could really also use some infomation on cross agent tooling connected to that.

by all means if you happen to run across anyone i have tons of questions. all good if not.

[flint vapor]

Yes the question you asking are always related to AI right?

[craggy slate]

@flint vapor yes. exclusivly

[flint vapor]

Ok ok, unfortunately is not my area about terraform and governance usage but for sure someone that knows more than me know about this😅

[flint vapor]

I recommend you to use deiteris fork

-wokada

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[flint vapor]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ First of all check your gpu and see if is compatible 😄

[craggy slate]

@flint vapor what is your area of focus pertaining to ai?

[quasi oxide]

this is old i think

[flint vapor]

RVC and i started learn for realtime voice changer AI

Also GPT-SoVits

[quasi oxide]

@flint vapor does ur pc have to be good to work well?

[flint vapor]

Mostly in this, then something few about llm models on ollama or lmstudio

[flint vapor]

I have a rtx 3060 is not so much strong but is ok for my voice changer 😄

Which gpu do you have?

[quasi oxide]

well its more for my friend

i cant use that

becuse of low gpu

[flint vapor]

Ohh do you know which GPU and OS your friend have?

[quasi oxide]

wait let me see

[quasi oxide]

gpu is rx 7600xt

and idk what os is

[flint vapor]

Oh for example like windows 11, windows 10 or a linux distro

Or Mac OS

[quasi oxide]

w11

[flint vapor]

Yes is very fine

-wokada

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[quasi oxide]

his voice changer sometime wont work

[flint vapor]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ Here there is a guide, say him to install the amd version 😄

[quasi oxide]

for some words

[flint vapor]

It depends from model mostly or from the chunks and extra he put

But mostly is from model

[quasi oxide]

can you spare some models?

Or do i have to find them in here

[flint vapor]

You can find it here or also in weights

[craggy slate]

@kakao im not familiar with using sovits, but the output, can that be built into a rag using payloading to give that voice change context from a source? i dont use that software - nor have i ever. I had to look it up to see what it was. my objective is to allow a part of my project more versatility. Could sovits be used to emulate lets say characters from lore? again, im ignorant on how it works. My second question - if im already training local models - could this be incorporated in that cycle?

[flint vapor]

On weights is easier to search than ehre

*here

[flint vapor]

You can use some scripts for make the character talk and then using an ai for example a local llm for give the ai the voice of your gpt so vits character, i don't know so much how to implement it but i know is possible merge an LLM with GPT-SoVits Voices

[craggy slate]

what if im not using GPT - could that be adjusted to a local model? could the GPU usage be offloaded to a ray node?

[flint vapor]

Oh GPT is the name of tool "GPT-SoVits" so is a unique script for AI TTS 😄

It can be confused with ChatGPT but is not

https://github.com/RVC-Boss/GPT-SoVITS

[craggy slate]

ill have to read up on it, for the cheat sheet, this can be coded into python? - diregard started reading the article thank you.

[flint vapor]

Yes you can code in python for use for example a local LLM with GPT-SoVits, i don't know if in the github page is explained how to do this but i know it can be done

[craggy slate]

thank you - i hope my previous request can also be assisted with in time. for now thank you.

[flint vapor]

You welcome 😄

[craggy slate]

Sorry, @flint vapor one more question, im reading the github, looking at some of things, for quick clarity - assuming hardware isnt a issue ( threadripper) can each agent using Sovits carry its own voice and talk to each other in a medium like discord if i already had the backend framework for that?

and i see it uses cuda, do you know if this causes issues with tensorflow?

[flint vapor]

It should not be in conflict with tensorflow, i don't know if more voices can be used in the same time, but for sure one at time yes, i don't know if all in the same moments can work together, of course you can try, but i never tried this😢 of GPT-SoVits i know how to fix common error mostly and how to set it up while you are in the interface

[craggy slate]

ok! is there a better software that you would recommend trying before i dive into this?

[flint vapor]

For TTS and AI is more personalizable GPT-SoVits 😄

[dense cairn]

Can someone send me a link for Apolo I also need help using it

[hallow thistle]

You mean Applio?

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[simple ore]

You're looking for TTS with voice cloning

there are many options based on the language

if you even need voice cloning

most of tts are done using pytorch, not tensorflow

[craggy slate]

@noobies thank you - im not entirely sure what im looking for. Voice implimentation only became a factor today. I am very open to any software I can introduce to my stack. Ive started implimenting sovits, but im ignorant of all voice implentations since i have no first hand exp with them. my stack has several multi function modules connected to faiis and interconencted rags through GCS. and terraform. my objective is voice input to actions with telemtry built. Is sovits the right software for that? or is designing my own a better investement? but thank you for the reply. if you have insight that is helpful to my objective I would love to learn. just @ me with any recommendations or articles you recommend if you choose to.

[simple ore]

gpt-sovits is a tts with voice cloning

it only supports a few languages

text to speech

[proven hill]

okay but what happens if i train an italian voice with the english pretrain? like it will sound THAT bad?

[simple ore]

technically the input voice is only used to calculate the conditional vector

in other TTS it may result in an accent

depends on whether tts was trained on a diverse set of speakers with varying accents, then it just catches that and generates tokens that match that accent

but it may just extract the timbre and speed

[proven hill]

gotcha, thanks

[simple ore]

with chatterbox

[proven hill]

jesus christ those are amazing

[simple ore]

same voice, just two different reference samples, thus different results with different accents and read speeds

[maiden wren]

Ok so what is best settings for it

For models

[low shard]

taking this for context: #🧬│ai-chat message

[low shard]

what tutorial link are you using? can you share a screenshot of ur wokada?

!give-media-perms 1h @maiden wren

[low shard]

also, what models are you looking for? for example roleplay as "peter griffin", or e-girl trolling/catfishing?

[analog obsidian]

sometimes i feel chatterbox is more accurate than rvc when cloning the voice
is it just me or it's just that good? lol
like, the voice resemblance is great

[maiden wren]

I want to use for myself

For test cus I never used one

[low shard]

yeah, I was asking are you gonna use it for roleplay/fun things or e-girl trolling in games/vc?

[maiden wren]

No, that's weird

[low shard]

so, no e-girl trolling right?

[simple ore]

not always and it depends on the voice

[maiden wren]

No trolling

[low shard]

great, catfishing isn't allowed

[maiden wren]

Okay

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ be sure to get wokada deiteris fork

uninstall everything you got off youtube

[maiden wren]

Okay

Also I have question can it work on my pc

Like will it lag

[hallow thistle]

If you think E-girl trolling thing is weird, it's okay and I agree. E-girl and catfishing are banned here.

[maiden wren]

Fr weird asf

[analog obsidian]

it's going to lag only if u play games with it and have a weak gpu

[maiden wren]

My gpu rtx 4080 btw ram 64gb because of after effects

[analog obsidian]

ah perfect, your gpu will be able to handle gaming just fine

[maiden wren]

Okay I hope so

[low shard]

it will work

[low shard]

also 64gb of ram are crazy lol

buy an rtx 5090 🔥

[proven hill]

i also have 64gb of ram

[low shard]

do yall really use all 64gbs

I find myself pretty fine with 32gb but i dont do work stuff like afet effects lol

[proven hill]

130 for 32 180 for 64, 50 euros for double the ram i can always use with flux is okay

[low shard]

ohh well nice then

[knotty moth]

128 GB upgrade plan would even be nice thing

[analog obsidian]

you need them for llm

flux also needs tons of ram

[low shard]

yea that's true

[knotty moth]

there are already some 2x64 GB ddr5 kits,

so prepare for the upgrade plan

[low shard]

I just got my pc 7-8 months ago 😭

[polar rock]

Someon send me colab link

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

@polar rock there's thousands of AI programs, also google colab is just a cloud computing service for people with a bad pc (or without one) to use a remote good pc

[hallow thistle]

Which Colab notebook? Is it RVC, W-Okada or Stable Diffusion?

[craggy slate]

@low shard if i provide a screenshot of my current software could you recommend a better version? in correlation to " This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program"

[low shard]

yes, but you'd need to elaborate also the rest

!give-media-perms 1h @craggy slate

[craggy slate]

so that i dont flood this chat, what specifically in terms of data, would you need, the screenshot i was going to provide is part of my live ecosystem and includes a AIMB - is that sufficient?

[low shard]

AIMB?

[craggy slate]

for context, i am currently building - every aspect of this - is there a software that expedites microservice, rag, docker, terraform, or other large scope aspects- i currently design and make my own testbeds in conjuction with pytest. and have started integration with voice, due to how i build in stages, modularly. I hope this suffices. the software im looking for are opensource or paid software that automizes this process.

and MCP server creation.

[craggy slate]

sorry, a ruleset for ai development-generated by ai. perhaps its called something different in your field, i wouldnt know - my objective, is to obtain opensource software or knowledge that expedites my project. Software that automates governance objectives with multi tool/function agents. preferable ones that deal with large pydantic 2 enums. and other DQN aspects.

[mental spruce]

Nick

I downloaded it but its like that

[low shard]

Sorry but I'm not aware about that subject, maybe someone like @simple ore can help you better

[low shard]

#✨│ai-help message btw would have been better if you make a #1192011222023950368 , making unorganized convos makes it harder for me to remember your help request

[mental spruce]

i just wanna download it...

[low shard]

could you please make an help post? it would be easier for us both since there's a lot of people here, and this is mostly for quick casual help, while yours is a bit more of a long convo

also it seems you downloaded the wrong one lol

please make an help post so its better for the both of us

[mental spruce]

done...

[simple ore]

nope, this is weird shit

[low shard]

I'm not really sure what they are talking about

[craggy slate]

hmm, sorry, there isnt really alot of documentation for what im doing.

or research. TLDR, im making a game engine that doubled as simulation software. built for persistent ai controlled governance. its nothing new. or novel. just silly and often difficult.

[tranquil wagon]

what happen to razor

yo

wheres razor

[viral mason]

He died

[maiden wren]

lowkey will go for rtx 6090 if it does release 😄

[dense cairn]

thank you

[low shard]

Lowkey get a 40k AI specialized GPU 🔥

[meager pollen]

Anyone know how to manually change chunk i cant get the exact settings i had bc of the sliders

[viral mason]

Use your arrow keys

After clicking

@meager pollen

[meager pollen]

Ty it works

[viral mason]

👍

[normal drift]

Guys, whats the best Colab notebook now for RVC AI stuff

like to input a song and make it sound like another character sang it with the character's AI model

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do

[low shard]

google colab is just a cloud computing service (remote good pc) with limited free time for people with a bad one

[normal drift]

Well since I'm looking for a Colab notebook, my PC's specs shouldn't be the issue

[low shard]

you need to first check ur pc gpu

[low shard]

it's better to do it locally first if possible, colab is unstable and has limited free random time

what's ur pc specs?

[normal drift]

Basically I have a laptop with 4GB RAM, without even a dedicated GPU, just "Intel UHD Graphics"

[low shard]

wow 😭

yeah that's cooked to another level

[normal drift]

Third world countey skill issue sorry

I mean I was looking for colab notebook for a reason

[low shard]

Train (make) RVC Models on cloud:

1. Prepare the Dataset
2. Setup RVC:
   Choose a cloud way to use RVC,

• Google Colabs (max 4 hours of daily T4 16gb gpu not granted for free, not much hours for training, but easy to use, there's a paid tier):
  • Applio (UI)
  • Mainline (U, broken right now)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus, either T4x2 16gb each or P100 16gb, only free):
  • Applio (UI)
  • Mainline (UI, broken right now)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly, Free Studios run 24/7 but require restart every 4 hours. There's a paid tier):
  • Applio (UI)
  • Mainline (UI, No guide as of right now)

3. Be sure to know about the tensorboard

Google Colab = Easier but risk of getting disconnected
Kaggle = Harder but way more gpu time
If you are looking for the easiest way and for free, is using https://weights.com/ which ofc uses RVC

RVC Inference (use models) on pre-recorded audio on Cloud

You can use either:

• Weights.com: Easiest Possible Ever Automatic
• Ilaria RVC Zero: Fastest free on cloud
• Applio UI Colab: RVC Fork with some extra features like TTS
• RVC AI Cover Maker UI: Automatically Separates the Vocals and Instrumentals, converts the voice and mixes them back

Gave you all the ones possible with a little explaination

[normal drift]

Thank you 🫡

[low shard]

You're welcome, let me know

[low shard]

I saw people with a 2k+$ pc before thinking colab was the only method, that's why we always ask the specs first

some have good pcs and think they cant run it, some other have a bad pc and think its good enough lol

[normal drift]

I see its understandable, I remember when Colab actively shuts down AI art gen notebook coz people abused it

I used to use Hina's AICoverGen notebook before it stopped working

[low shard]

AICoverMaker Web UI is better than that btw

[normal drift]

Btw is there a good Colab notebook for AI art stuffs too

[low shard]

yeah, i mean overall cloud is more unstable, bc google could update dependencies which break some notebooks, and if the tunnels API is down, you can't access the WebUI

[normal drift]

I remember when Automatic1111 UI was the meta

And I mainly used some notebook from the Waifu Diffusion server... then Colab itself shuts it down

[low shard]

https://colab.research.google.com/github/cpacker/stable-diffusion/blob/interactive-notebook/scripts/stable_diffusion_interactive_colab.ipynb, but im not sure if this even works, would be better to use a service like https://voyages.weights.com or https://civitai.com since its very risky to use colab for ai art

[low shard]

Waifu? 😭

Just so you know generating anything NSFW is just going to increase your ban chances

[normal drift]

I don't generate anything with stuffs like nipples or genitia, just getting idea for OCs or anime chars in different outfits ... Btw Waifu Diffusion was a finetuned model from Stable Diffusion for anime stuffs

Also with those specs I have no choice but to be beggar when it comes to AI gen xd

[low shard]

anyways, if you do it on colab, it's prob best to do it on paid tier since on free tier you could risk ban

tbh id just use weights' voyages

ofc you can't expect unlimited free stuff since they paying for the server, but better than nothing

[normal drift]

Thanks, I didnt know about that one b4

[low shard]

it would just be better you save money and do it locally

you're going to have limitations on free cloud

[normal drift]

I'd like to if possible, but dont worry I'm not some sort of content creator, sometimes I randomly need OC ideas for drawing I just want a few quick AI gen shots to materialize the intrusive thoughts XD

[pine star]

What's meta now?

I'm still have automatic1111 installed rn, but that was from a long time ago

[maiden wren]

Probably using old version from method of YouTube, go to guides tap click pinned guide which will take you to another channel where ai hub link is and click on link and read deiteris' W-okada guide and disable firewall that's what I did to run it

[simple ore]

comfyui if you're into drawing complex workflows, SD.Next if not

[azure patio]

do i run this cell in end or how? i wanna backup later to continue training my voice model, so how should i proceed?

[flint vapor]

Download the deiteris fork, the one you trying now is outdated

I give you the guide

First of all check the gpu you have and see which version download and under you have the charts

-wokada

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[flint vapor]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/

[rain hare]

Does running the real time ai voice changer affect gpu cus im using a shitbox rx 560 XT

[toxic plinth]

sup y’all I need a help with Weights Ai.So I want to add a cover to a community server and idk how to put it in there can someone help with it?

[low shard]

Rx560? I think that's not even supported lol

[rain hare]

But will it like take alot of power

Oh

[low shard]

I misread it at first lol

[rain hare]

So should i not try and jus delete it

[low shard]

Do you have an RX 560 xt, or an RX 5600 XT?

[rain hare]

RX 560 XT

The 8 gb version

If you say it works after i deleted it

I spent 3 hours downloading it

[low shard]

Ehhhhhh, you could try, be sure to use https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/, don't expect to be using it in ANY games, just VCs

It will for sure not have the best performance, but it's just a try

[low shard]

If you followed video tuts, it was just wasted time btw

Because there's no updated video tut

[low shard]

It would be better to ask in discord.gg/weights

Will help in https://discord.com/channels/1159260121998827560/1394479814693490768

[simple ore]

"load a backup" copies the backup back into colab

[acoustic lion]

I was doing the realism test, now I try to test how I sound and it's just endlessly echoing.

[acoustic lion]

@low shard

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

be sure u aren't using yt tuts for realtime voice changer, never ever trust them, all old

yes its outdated asf #🧬│ai-chat message, that's an extremely old version of original wokada, and vb audio cable cause issues on windows

Elaborate:

• your pc gpu
• operative system
• what you want to do

then you got an extremely old version like the other guy, elaborate what I asked you first #✨│ai-help message

[low shard]

realism test?

[acoustic lion]

https://docs.aihub.gg/rvc-voice-changer/realism/

@low shard this section here, I was a little confused by the layout of Voicemeeter since it didn't have b1b2b3 etc, it just had A or B (I think I got it right, unsure)

I then installed light host and got the VST's (the one that were suggested) and when i go on discord to try to hear it, I just hear multiple echoing

[acoustic lion]

Rather I meant, testing out the realism section

tuned them exactly to the pics, whenever I talk I just hear like 3 of me talking

i don't have the same layout as the tutorial shown in voicemeeter too

[low shard]

I didn't work on that part of the guide, it's better to ask like @analog obsidian

[acoustic lion]

oh okay

hopefully they see your ping and respond

[analog obsidian]

i dont use any of those things

[acoustic lion]

They useless?

[analog obsidian]

yes because a good model can sound realistic without filters

[acoustic lion]

this is a real time voice changer?

[analog obsidian]

yeah a rvc model i made last day

[acoustic lion]

Wow, kinda sounds amazing

[analog obsidian]

there is some ways to increase the realism using some filters but i dont want to say what

[acoustic lion]

How come?

[analog obsidian]

[proven hill]

500 dollars paypal to lyery and its yours my friend

[acoustic lion]

damn

[proven hill]

https://tenor.com/view/laughing-duck-gif-1363576254939741235

[analog obsidian]

[acoustic lion]

Then err...

Who can exactly help me then with the light host thing? I only have one filter active and my voice is just echoing

[analog obsidian]

it does echo without light host?

[acoustic lion]

Nope

With it it does

[analog obsidian]

hmm, huh, weird
might be some driver interaction, have you tried enabling "echo" in w-okada?

[acoustic lion]

I tried that yeah

Seems like everytime I add a VST the echoing increases

as if every VST is just sending out it's own voice sample

[analog obsidian]

thats weird... the member who made the realism section is banned sooo

he was the only one who knew how to fix those issues

[acoustic lion]

Oh.

[analog obsidian]

☠️

[acoustic lion]

Uhhh, weird request, but would you be up to get in to a voice chat at least to hear how mine is currently sounding? I'd like your input on it and see if there's any suggestion you can make towards how I speak

without the echoing of course

[low shard]

it was a joke btw lol, we don't do paid comms in the server

[acoustic lion]

Oh I know that no worries

[analog obsidian]

sorry im not a native english speaker

[acoustic lion]

I mean... neither am I

[analog obsidian]

can you send the audio here instead?

[acoustic lion]

is it okay if I dm you it?

[cold shoal]

Hello can you help me again

[proven hill]

who?

[analog obsidian]

i don't accept dms from this server

[analog obsidian]

razer

[acoustic lion]

oof

[proven hill]

razer is banned? the fuck? for what lmao

[analog obsidian]

a veeery meaningless and stupid staff drama 😭

[acoustic lion]

can yall bring him back for like 2 minutes

[proven hill]

https://tenor.com/view/raised-eyebrow-eyebrow-raised-eyebrow-emoji-silly-cat-wtf-gif-8109444744753568214

[acoustic lion]

just so they can tell me how to fix it lmao

[low shard]

Great enough pc, I'm guessing windows 11

I'm guessing you followed that novision guy on youtube? you should uninstall vb audio cable from windows app settings

[analog obsidian]

i can ask him

[acoustic lion]

That'd be nice ^v^

[analog obsidian]

done, now we wait

[low shard]

It wasn't stupid staff drama, and you know the real reason, also you shouldn't be talking publicly about this stuff

[analog obsidian]

fine then

i wont say anything else

[analog obsidian]

he said "He prob has the output set to the wrong device"

[acoustic lion]

let me see

in light host it self i assume

[proven hill]

its discord 💔

[acoustic lion]

don't think it's the wrong one @analog obsidian

[low shard]

yeah I know that guy, his outdated tutorial tormented us, making a 1 month old tutorial with info from 2023

be sure to also delete the wokada .zip and folder

[analog obsidian]

k i told him

[acoustic lion]

lyery being a goat rn

[proven hill]

catfisher

[acoustic lion]

one day i'll see someone say " Roleplay "

[analog obsidian]

i just use models of game characters when i play with my friends

[maiden wren]

Blud got banned

[hallow thistle]

Yes, catfishing is banned in this server. Just letting you know you should never play with anyone here.

[acoustic lion]

I mean... yeah

Catfishing in general is just a no

Not to mention everyone became aware of AI models mainly due to catfishing videos

[low shard]

actually RVC became very famous by AI Covers, this server was originally a kanye west ai cover server

[acoustic lion]

Didn't yall say RVC doesn't mean Realtime Voice Changer

[low shard]

yeah it doesn't

RVC means Retrieval-based-Voice-Conversion

[acoustic lion]

sooooo, techincally I am right

i spoke of ai for catfishing, not RVC

Which granted both are kind of the same thing at the end of the day

[acoustic lion]

no wonder your model sounded so good in quality, you probably fed it hella audio

[low shard]

https://tenor.com/view/absolute-cinema-ultrakill-gif-v1-v2-gif-877973826586985475

[analog obsidian]

i used 3 hours of data

[acoustic lion]

DAMN

Yeah hmm

No luck on my end regarding the filter though

[analog obsidian]

yea im sorry... razer is gone and he is busy atm

oh wait

he answered now

xD

[acoustic lion]

oh

w

[analog obsidian]

So he has in voicemeeter stereo input 1 set to line 1 and in light host he has his input as line 1 and output as voicemeeter input?

he said that

[acoustic lion]

Yeah that's... everything I have

i relaunched okada fork and now i hear myself twice

@analog obsidian oh yeah for some odd reason, sometimes after using okada fork for a bit, the voice becomes weird

and the only way to fix it is by relaunching it

[maiden wren]

Doesn't happen to me lol

[analog obsidian]

huh weird

[analog obsidian]

what if you try vonovox instead?

[analog obsidian]

he said everything its working for him and that you probably got something wrong

[acoustic lion]

is it better?

[analog obsidian]

yea but lacks some minimal features, like the monitor one

[acoustic lion]

Hmm don't really care about that feature...

But can you explain how better it exactly is?

[analog obsidian]

it is a new program with a totally new code and much better optimized, in theory, it should be faster than wokada.

[acoustic lion]

And the quality?

It's pretty much my main concern

[analog obsidian]

the same, you cant change that unless you update rvc2 to rvc3

[acoustic lion]

is dat even possible

[analog obsidian]

maybe no lol, devs tried it but failed

[acoustic lion]

frickity

[low shard]

haven't tried vonovox yet but it has the fp32 settings that helps with quality/stability?

[analog obsidian]

uses tf32 inference

[acoustic lion]

team fortress 32 holy

[analog obsidian]

are your wokada settings right? razer told me maybe u got the input and output wrong in the app

[acoustic lion]

nah

my mic as the input

and line 1 as the output

i read the entire section

[analog obsidian]

damn

[acoustic lion]

did u show him the voicemeeter thing?

i don't have b1b2b3 like he does

nor this

@analog obsidian

[analog obsidian]

he said u might need voicemeeter banana

[low shard]

Hello, elaborate your help, sorry but I don't really remember you

[acoustic lion]

WHAT

isn't that paid?

[cold shoal]

Um so I wanted to know if I could get AI models made for me

[acoustic lion]

why does it say i already have it...

because i doooo

x-x

[analog obsidian]

😭 ok i can make an exception
imma dm u so you can get in contact with him

[low shard]

You can search rvc ai voice models at:

• https://discord.com/channels/1159260121998827560/1175430844685484042
• In https://discord.com/channels/1159260121998827560/1163592055830880266 , Do /find with @earnest musk
• https://weights.com/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• Suggested Models for Realtime Voice Changing (Wokada)
• 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:

• make it yourself with our docs guides
• Ask a free request in https://discord.com/channels/1159260121998827560/1159289738314919936
• Be aware that we don't allow any paid comms in the server

[earnest musk]

:wave: @low shard, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models

[analog obsidian]

razer is literally the only one who knows how to fix things related to the realism section

[low shard]

Were you looking for me to train a model for you? Sorry but I don't really do that, but I linked you a place where to ask free requests and look if the model was alrwady made

[cold shoal]

Also how do I record ai songs

[low shard]

wdym?

please elaborate ur pc gpu, operative system, and what you want to do

[acoustic lion]

he wants to do ai song covers

[cold shoal]

Exactly

[low shard]

elaborate ur pc gpu and operative system

[analog obsidian]

check dms

[cold shoal]

Wdym

[low shard]

tell your pc gpu and what operative system do you use

that's crucial information for AI

[cold shoal]

I have a MacBook if that’s what you’re talking about I just don’t understand

[low shard]

macs aren't the best for AI,

For Inference (use models) Mac, You can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides, probably won't be able to train, make models):
  • Applio: A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
  • Mainline: The original RVC
• Cloud (remote good pc, easier and faster than ur PC but limited time):
  • Weights.com: Partnered with AI Hub, lets u do them easily but u may be in a queue
  • Ilaria RVC Zero: fastest and simplest that you can get for free
  • Applio Colab: max 4 hours, not granted, of GPU
  • RVC-AI-Cover-Maker-U Colab & Kaggle: Automatically separates the vocals and instrumentals, converts the voice and mix all together back

Easiest possible (automatically separates vocals & instrumentals) : weights.com & rvc-ai-cover-maker-ui colab/kaggle
Easiest cloud: Ilaria rvc zero
Easiest Local: Applio

[hallow thistle]

Apple Macbook is not a GPU, it's a laptop. Does your Macbook have Intel or M-chip (Apple Silicon)?

[cold shoal]

I don’t think so

I can take a look if you want me to

[low shard]

your mac is either an old intel or m chip one, there isn't any other option lol

[low shard]

you should prob look into the system info

this is open source AI, it runs directly on your hardare, it isn't a product like chatgpt lol

[hallow thistle]

Dedicated GPU like NVIDIA GeForce RTX or AMD Radeon RX would unlikely to be in a Mac, so any AI program would run on CPU instead, regardless if your Macbook has Intel CPU or Apple Silicon chip.

[gentle shell]

Hello, I want to try and get my voice to sound like an alien, how can I do this?

[sharp glacier]

Can someone link the download link for the voice changer and only that I keep finding stuff that I think is it but isnn't
And when a page lists 22 dif downloads it doesn't help my brain process

[flint vapor]

-wokada

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[flint vapor]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ Here you find a guide 😄

Which gpu do you have?

[sharp glacier]

that's the issue I read it all and still don't understand what link I download.

[flint vapor]

Which gpu do you have? I will tell you which 😄

[sharp glacier]

I just want a direct link to the only thing I need.

Idk how to check gpu all I can tell you is it has graphics. and is a Nitro 5

[flint vapor]

Oh yes i will send, i need only which gpu you have for see which right version give you

[knotty moth]

find some https://discord.com/channels/1159260121998827560/1175430844685484042 sounding like that

[flint vapor]

Go on Task Manager

[sharp glacier]

idk where that is

[flint vapor]

Performance

[sharp glacier]

i'm old..

wheres that.

[flint vapor]

Right click the windows bare under

where you have all apps

Or you can also search on windows search, task manager

[gentle shell]

Yeah, but how do I like use the models?

[sharp glacier]

idk what to click

[proven hill]

there

[flint vapor]

Yes right click the white space

Right on roblox

[gentle shell]

Are there like any tutorials?

[knotty moth]

you should look at your Nvidia/AMD gpu tho, not the igpu

[flint vapor]

There is AI Hub docs website i send you

[flint vapor]

https://docs.aihub.gg/

[sharp glacier]

i'm not pronounceing / retypeing all this...

so RTX something

[proven hill]

model

[sharp glacier]

I can't take a screenshot I tried. but it's RTX something..

[flint vapor]

Do you went in task manager?

[proven hill]

we need the name of the gpu

[sharp glacier]

it's Nvida RTX something.

[flint vapor]

If you have windows 11 click here up left on your task manager

[proven hill]

youre trolling

[knotty moth]

ye it's the Nvidia gpu

[proven hill]

NEXT

[flint vapor]

and select performance

[sharp glacier]

intel R uhd something.

[proven hill]

so you lied

[flint vapor]

It wasn't rtx before?

[knotty moth]

https://tenor.com/view/sealyx-naked-gun-the-naked-gun-facepalm-facepalm-meme-gif-11743768702765397589

[sharp glacier]

it was.

it lists 2

[flint vapor]

The important is the rtx one

[proven hill]

whats the name of the rtx one

[sharp glacier]

idk what to tell you.

[flint vapor]

Probably you have hdmi connected on motherboard and not gpu

But you can use your rtx same

[flint vapor]

Go on here

These three lines

When you click on task manager

or just do ctrl alt canc

and you can select task manager too

[sharp glacier]

-this.

[proven hill]

tell

us

the

name

of

the

[flint vapor]

Yes but we have to understand if is a series 5000 or lower

And also for help you set settings for the voice changer

[gentle shell]

Thanks goat

[sharp glacier]

I already said the name

It's RTX nivada something

[flint vapor]

Yes but there is a number near

[flint vapor]

https://www.youtube.com/shorts/wVvN8x8roHY?feature=share

Try like this

Go on device manager

Display adapters

[sharp glacier]

I doug in my trash And find the one I use to use if it works yay if not oh well.

I wasn't going to go through 20 something dowload links take screenshots over and over just to get to it sorry.

[flint vapor]

You can send me screenshot in dm if you want don't worry

[sharp glacier]

if it doesn't work I will

[flint vapor]

Sorry if we always ask,is for help you and make you have best performance for voice changer

[proven hill]

what do you need it for anyway

[gentle shell]

Managed to get it running, thanks again!

[flint vapor]

You welcome! Which tool you want install? or you using?

[gentle shell]

Deiteris' W Okada Fork I think

[delicate beacon]

Hi, I'm running the w-okada fork on Kaggle and it has been working fine for the last 24 hours, but I am now suddenly getting this error in the kaggle console:

Faiss assertion 'err__ == cudaSuccess' failed in void faiss::gpu::streamWaitBase(const L1&, const L2&) [with L1 = std::initializer_list<CUstream_st*>; L2 = std::initializer_list<CUstream_st*>] at /project/faiss/faiss/gpu/utils/DeviceUtils.h:163; details: CUDA error 33 invalid resource handle

And getting the error '[SIO] rconnection failed Error: xhr poll error' on the ngrok site itself

Please could someone help?

[flint vapor]

Have you used the kaggle that there is on ai hub docs ?

[delicate beacon]

Yeah the deiteris fork

@flint vapor

[flint vapor]

I’m not so expert in cloud but have you tried reboot the kaggle? Or maybe did you finish the limit ?

[delicate beacon]

It says I've only used 5 hours out of the 30 available?

[flint vapor]

Have you try reboot the kaggle right ?

[delicate beacon]

By rebooting do you mean factory resetting?

I've tried to terminate the session and restart it many times

[flint vapor]

https://docs.aihub.gg/rvc-voice-changer/cloud/w-okada-kaggle/ I recommend you to follow the guide again

[split iris]

yo, anyone who uses wokada, it said localhost couldnt be verified but like...is it that deep?

[narrow berry]

hey could someone pls help me what tool is used for this ive actually spent weeks trying to figure it out and tried loads https://youtu.be/ZPhzReziYN8

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[elder knot]

Im a bit new. What does "Epochs" means?

[simple ore]

how many training loops the model done around the dataset

more does not mean better

[elder knot]

Oh, i see

ty

[vale osprey]

-colab

[patent trellis]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[viral mason]

what's the best models for echo and reverb removal in the uvr huggingface space

[storm fern]

guys i need help

for some reason when i install the virtual audio cable it doesent make the ai voice work

[storm fern]

do you know how

[viral mason]

you'd have to elaborate more

[storm fern]

so when am trying to use the guide

am trying to install the virtual cabels

but when i installed them

it wont work

i dont know how to make it work so it plays back my ai voice thingy

[sharp glacier]

what am I doing wrong whenever I use it it has a awful Echo I can't get rid of it I have tried turning the echo box on have tried turning sup and sup 2 on diddn't fix turned em off still their idk what will stop the echo I even turned my main pc volume to 0 still occurs.

[knotty moth]

• try using headphones that don't leak to the mic
• if you're using some audio routing tool like voicemeeter, check the configurations

[sharp glacier]

the issue is I don't have any other headphones and have 0 funds to buy any.

as far as the 2nd issue I voice metter banana. and don't fiddle with settings

this is what it's set up as.

I don't use regular vm beacuse it's to confuseing and complex.

and I perfer simple vs 3 day setup/getting mad.

but i don't have it running anways

I use virtual audio cable only. That's all I use with the voice changer..

[knotty moth]

there are good headphones not above 100 bucks

or even some earphones for few bucks

[sharp glacier]

I can't afford any I have the headphones I use rn which is Xbox headset/pc it works fine, for everything else so I see no reason to upgrade..

they ran me 40$ 2 years ago and work perfectly fine.

[knotty moth]

@slender python if you're looking for the best prompt adherence, it may be SD 3.5

otherwise you may need some guide to improve your prompting like this

https://education.civitai.com/civitais-prompt-crafting-guide-part-1-basics/

[viral mason]

any mod know?

[digital vapor]

Hi, what does this mean? CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Also it happens when i try to start my model.
And it spams "An error occurred during voice conversion. Check command line window for more details."

It doesnt give the error, when i restart the whole voice changer.

It works completely fine, but i dont understand why did it do that.

[storm fern]

could someone pls help me with my problem

[simple ore]

what's your GPU?

[upbeat cairn]

how do i download the client ?

[clear citrus]

where here? download
?

Where i can find deiteries download?

[proud nebula]

How do i use the okada voice changer software with an amd card?

[pastel oak]

Whats the full name of your amd gpu

Depends on your gpu though

[flint vapor]

In the link that shady sent you 😄

[knotty moth]

something that could be sought in this server

[flint vapor]

Which client you mean?

[slim schooner]

Hey guys, so i managed to clone a voice. However, it still sounds a bit robotic at very random times. I trained it up to 150 epochs but i feel like i should train it a bit more. My solutions for a better clone are 3 things: more training time, selecting a better pretrained model, and selecting the right embedder model.

I'd like to note that the current cloned voice sounds good but the robotic effect is what really degrades it. To solve the robotic effect issue should I do all 3 things I mentioned above? It's a 50 min audio data set that I'm working with. The audio is cleaned and free from any background noise.

I'd also like to note that I'm cloning a voice to speak in Spanish. What embedder and pretrained model should I use in this case? I used TITAN pretrained model which came out good so I may keep that. Should I also choose a custom embedder model?

Any help is appreciated!

I had sent this message a while ago but I guess it was overseen or missed. Just need some guidance.

Or if anyone who has had success with cloning voices multiple times, if they could share their settings. Cause I'm not sure what I'm doing wrong lmao.

[vast elm]

i'm not exactly sure what would be the best settings to use for a spanish voice model as i've only trained english models personally, but wanted to ask, the 50 minute dataset you are talking about is a huge amount, likely more than you need to create a very high quality voice so im suspecting maybe the quality of the dataset is not as good? when training a voice model to sound natural the quality of the dataset matters much more than the quantity

take some time checking this guide out on how to prepare a high quality dataset
https://docs.aihub.gg/rvc/resources/dataset-isolation/

[slim schooner]

50 mins is a lot? I remember talking to someone and they said that they used about 45 mins or so of data to clone a voice. The quality of the data set is pretty good i'd say. It is taken from the same source used with a high quality microphone, any background noise was removed/cleaned using UVR.

I'll check out that link though. Thanks 👍

@vast elm Hmm ok, there are some things I need to redo then. Thanks for sharing that link.

[vast elm]

Of course, if you needed any help within any of the steps let me know i'll be glad to help

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[low shard]

Ask their GPU first, saw 2 people with Intel UHD earlier

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

be sure to not use YouTube video tuts, they all old

[vast elm]

@low shard you gotta chill out with the same copy pastes

like one guy asked "where can i find deiteries download" you didnt even need him to elaborate, he just asked for the link 😭

[low shard]

2 guys yesterday did the same thing, then complained that they couldn't run it because they got Intel uhd

You have to always ask to elaborate and know the user's PC GPU & OS

Else they will just waste time trying to run things on 2000s hardware, someone also tried using it on Intel(R) G41 Express Chipset from 2009 💀

[low shard]

Along with someone that spent 1 hour trying to install wokada, then realized what they wanted was ai covers, not a realtime voice changers

You have to be careful and ask the user to elaborate else you will waste time, because tech knowledge is not common

Then there's people with an RTX 40 serie asking for colab lmao

[vast elm]

crazy 💀

[low shard]

Actually not crazy, this happens frequently, the least a user elaborates and the higher chances the entire thing will be wasted time trying to do it asap

Oh btw I just noticed you didn't apply for helper?

[tender hedge]

hello

[low shard]

Hello there, do you need any help ?

[tender hedge]

is there any app for a realtime voice changer

?

[low shard]

Yes, this is a general ai server, but you have to elaborate first to check if your PC is good enough

Elaborate:

• your PC GPU
• your operative system
• what you want to do

[low shard]

Also, if you don't know your PC GPU

You can check your pc gpu on Windows via:

ctrl+shift+esc (task manager) -> Performance tab -> GPU

[tender hedge]

hmm i have an rx 6600 and windows 11 i want to use ai voice realtime to talk to someone

[low shard]

Ohh AMD, are you going to use it in games or discord VC? Or for e girl trolling/catfishing in games/bc?

[tender hedge]

yes bro hurry

[low shard]

You're going to use it for e girl trolling/catfishing?

[tender hedge]

noo i use it for playing with my friend

[low shard]

Alright

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

1st link

Wokada deiteris fork, read it

[low shard]

You will have to play on lowest graphics, 1080p with 60fps cap btw

Because ai is intensive and complex

Especially realtime

So don't expect to play super intensive games, especially on good graphics

[tender hedge]

okay is this free right?

[low shard]

Yes, don't trust YouTube tuts about realtime voice changers, they are either old or suggest paywalled things

[tender hedge]

give me a guide plsss i cant quite do this

[low shard]

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

1st link

[tender hedge]

then?

[low shard]

You have to read it, there's no video tutorial

[tender hedge]

okay after i download it what should i do i got the zip file

[low shard]

Did you get vac lite via https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable ?

You have to read it carefully, because missing a step can quite fuck up your whole audio system

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[tender hedge]

done

[low shard]

Alright, read the guide and let me know for any issues

[tender hedge]

ok done installing

now what now???

[low shard]

Did you set your usual devices for playback and recording tab?

After using setup64.exe

[tender hedge]

yeah i set it to my default speaker

[low shard]

Alright, extract the wokada zip, then go in the folder and open the .exe

[tender hedge]

and may i know the folder name?

hello?

i cant not see the exe files bro

[latent kettle]

Something like MMVCServerSIO

[tender hedge]

its Spec file not exe

ok i will look at the readme text

[low shard]

MMVCServerSIO

It's all written in the guide

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#opening-on-windows

[tender hedge]

opinion on voice.ai app?

[flint vapor]

I recommend you to use VAC

You can find it on deiteris guide

If you using deiteris, obviously

If you don't have deiteris fork i recommend you to download

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#opening-on-windows here you can find all

Is like VB Virtual Audio Cable, but i find it better and reduce the crackling

So i suggest you to uninstall vb virtual audio cable and install vac, you can find it on the guide 😄

[tender hedge]

uhm i already install the vac and i cant seem to know how to use it in discord when i change the input to that it wont work

[tender hedge]

after i install everything it seems like the voice changer its working but when i testt it in discord the vac dont have any sound

[flint vapor]

Is all in the guide

[flint vapor]

After installing the Virtual Cable, it changes your default audio system. Click Yes when it asks you to open the audio device settings (or press WIN+R, type "mmsys.cpl" if you closed it already), and change your Recording and Playback devices back to your usual devices. Same for communications device aswell (right click -> set as default communication device)

[tender hedge]

ty bro i think i missed the recording thingy

[sterile needle]

hello why is my voice shaking and glitching any solutions???

[maiden wren]

Increase Chunk

And seconds to higher

[short oracle]

is there a way to make it

have less delay?

[slim schooner]

alright im currently cleaning the audio with UVR, do you think you can share the settings you apply in applio? like screen shots? do you use a custom pretrain and embedder? i know mine will have to be different but i want to try it with the ones you have.

[vast elm]

well you preferabely need a pretrain which a good part of its training data would be spanish or languages with similar accent, as i said i have only trained voice models in english so im not very sure what would work the best for your scenario

but don't worry about the pretrain and embedder that much, the model sounding "robotic" is most likely the cause of a low quality dataset

if you manage to prepare a decent dataset i can promise you the voice is going to sound alot sharper and higher quality

[low shard]

Sucks ASF, same technology but you have to pay and stays in 24/7 in background to use it to power other users

Vb audio cable is bad, what's your PC GPU? Did you use a YouTube video tut?

[slim schooner]

yeah it was my bad because i initially screen recorded which made it into an MP4, then i converted that to wav and then I did a bunch of conversions that prob tanked the quality, but to the human ear it sounds fine but the AI prob sees those faults I suppose?

the TITAN pretrain model apparently is good for spanish so im keeping that. hopefully it turns out well this time.

[low shard]

Do you need any other help?

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[sterile needle]

rtx 3060 Ti
windows 11
i want to not sound like a robot
https://www.youtube.com/watch?v=SxdnGxicJOg
i cant screenshot here

[slim schooner]

lmaooo poor Nick getting bombarded with questions

[low shard]

That tutorial is outdated ASF, it uses an over year old version of original wpkada, and vb audio cable causes issues on windows, uninstall and forget everything

All video tutorials are outdated for voice changers

[low shard]

Also, are you looking to do e girl trolling/catfishing like duckus and novision?

[sterile needle]

what should i use then

[sterile needle]

nah

[low shard]

Are you trans? Since I saw that was a tutorial for e girl

[sterile needle]

bro i found that vid so i used it im not into that stuff

[low shard]

Great then

[low shard]

First, delete the wpkada folder and zip

Then, uninstall vb audio cable from the windows app settings, not just the folder and zip

[sterile needle]

sure

[low shard]

Then

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

read wokada deiteris fork

There's not any updated video tutorial, and you should never trust them for realtime voice changers as there will 99.9% never be any

Videos get outdated easily, ai moves fast

[viral mason]

I make good tutorials

[low shard]

there is no single, and there will never be, an updated vide tutorial for realtime voice changers, you can uninstall everything

Also, open source ai is intensive and complex, you need a good pc

Your pc is too Weak for Wokada locally, You got 3 options:

• Buy a better pc
• Run it locally (on ur pc) using the CPU mode of the wokada fork which has better performance https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ (but this isn't suggested as it could be unstable)
• 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) :
  • W-Okada's Deiteris' Fork Voice Changer Google Colab (currently works only on google colab PAID tier)
  • How to use Original W-Okada's Voice Changer Google Colab (has a Guide) (currently broken)
• Kaggles (30 hours weekly of better GPUs, T4x2 & P100, harder to use, requires an account and a phone number):
  • W-Okada's Deiteris' Fork Voice Changer Kaggle (the best and only working one currently for free)
  • Original W-Okada's Voice Changer Kaggle (currently broken)

[maiden wren]

Get better pc

[slim schooner]

ok so im probably gonna keep TITAN pretrain model, but can you share the other settings? like what you usually use? ill tweak the other stuff in regards to my gpu

[low shard]

@spiral sedge let me know if you're going to buy a new pc, or use wokada deiteris fork kaggle cloud

[analog obsidian]

the pretrains language doesnt matter that much

finetuning overwrites pretty much every previously learn phoneme
the reason why some non english models have bad pronunciation is because contentvec was trained mostly on english

[maiden wren]

Turn up chunk

Then turn up seconds

And also

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[maiden wren]

Turn up the pitch and lower the setting below it if that doesn't work then it's model that isn't right or you just don't have voice deep enough

[low shard]

this is the help text channel, you just need to elaborate more

[slim schooner]

what would be the best pretrain model? im using TITAN but not sure if its doing anything tbh

[analog obsidian]

original pretrain

[slim schooner]

the one that comes with applio?

[analog obsidian]

yup, its the best we have

[maiden wren]

Yo what happend to that guy's message

[slim schooner]

alright im gonna run some training.

[analog obsidian]

titan is a finetune of the og pretrain, so calling it a pretrain is wrong

[slim schooner]

also quick question, what is meant by batch size? bit confused, does less mean better results but longer training time?

[analog obsidian]

number of files ai learns every step

if u got 10 mins and above you can set this max to 64

just go the highest u can

for less than 10 mins i would not go over 8

64 batch size = the model learns 3 minutes per step

[slim schooner]

i got like an hour of data, but isnt this in regards to the available vram? my rtx 4070 has 12gb so i should keep it at or less than 12 no?

ok so go all in

[analog obsidian]

hmm im not sure if 12 is going to fit that vram

if you notice your training is very slow is because you got oom

but it doesnt hurt to try

[slim schooner]

ok so you would go all the way to 50 if you had 50 gb of vram? im so confused lmao

[analog obsidian]

the description of the batch size in applio is wrong, ignore it

i told noobies to change it last time

[slim schooner]

ok so ignore that, for my 12gb of vram, what should i select?

[analog obsidian]

maaaybe batch 16

or even less

if oom try batch 12-13-14

[slim schooner]

i tried 12 last time and my pc was struggling tbh lmaoo, ill try again but its basically gonna be using all of my gpu. ill try 16 but prob gonna need to close everything and just let it run

[analog obsidian]

you can try checkpointing

slows training by maybe 30s?
but allows for bigger batch

[slim schooner]

ok ill do that

how many epochs you think i should train to? i know i should use the tensorboard but it usually stays the same.

it never drops below 4.2

[analog obsidian]

until the model sounds metallic and bad

yea the graphs are kinda useless

only useful are mel and the grad g norm one

[slim schooner]

ok ill prob train to 250, thanks lyery

[analog obsidian]

[slim schooner]

should i cache data in gpu or nah?

[young halo]

Can someone show me a tutorial of how to resume training on applio colab? I always get an error when i try and it won't show any gradio link

[analog obsidian]

nooo

dont do that
it will eat your vram

lmao

[slim schooner]

ahhh thats what i did last time

also for 1 hour, what index algo should i choose?

[analog obsidian]

auto

[slim schooner]

leave it like this right? so applio uses the original?

[analog obsidian]

yep

[slim schooner]

audio cutting to simple right? or skip?

[analog obsidian]

i use auto

[slim schooner]

silent training files? default?

[analog obsidian]

2

[low shard]

dont eat this poor man vram

[analog obsidian]

[slim schooner]

alr thanks, ill let it train 👍

[young halo]

pls help me

[analog obsidian]

great

[low shard]

@oak plank did you setup dank memer to work only in #🤖│bots ?

[slim schooner]

lmao last question sorry, process effects checked or nah?

[low shard]

https://docs.aihub.gg/rvc/cloud/applio-colab/#section-46

[analog obsidian]

yes

[low shard]

There's nothing else, the one you got is worse and won't work anyways, you can't do anything other than buying a new PC or use cloud

All other programs like voice.ai are paywalled and worse, for example voice.ai makes you pay and stays 24/7 in the background to use your PC GPU power to power their services for other users, and other programs also use paid cloud

AI is intensive and complex, you can't expect such weak hardware to work

[viral mason]

it's payed

free version sucks

[low shard]

paywalled too, the free versions is just 3 daily random voice effects that sucks, I can say from personal experience lmao

as i said, your only option is either buy a new pc, use the free kaggle cloud wokada deiteris fork, or buy for a scummy service lol

[young halo]

do i have to input the dataset again?

[low shard]

to help you in case you might be confusing programs

[low shard]

you shouldn't have to, but use the same settings

[young halo]

Yeah but it's asking me for one, it doesn't show the same that l submitted even when i resumed successfully

Oh

I have to skip the preprocess part, Right?

[viral mason]

[low shard]

yes

[low shard]

[viral mason]

cat

[young halo]

Now it's working just fine, thanks!

Finally learned how to use applio for training lol

[low shard]

You're welcome

[mellow notch]

Hello, can anyone recommend me a good voice changer (my laptop is not that good)

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do

[low shard]

also, be sure to never use youtube video tuts for realtime voice changers, they are all old

and don't use software like voice.ai, it's the same thing but paywalled and running on the background 24/7

[mellow notch]

PC specs,
i5-10th gen
Intel arc graphics
12gigs of ram
runs on win11
Im creating my own anime series (1st one ever) and i wanna act out the voices and not get my freinds involved lol

[low shard]

Intel arc graphics
Are you talking about a dedicated gpu like intel arc b580 ?

[mellow notch]

nah integrated

[low shard]

Im creating my own anime series (1st one ever) and i wanna act out the voices and not get my freinds involved lol
RVC = Retrieval-based-Voice-Conversion, the best Few Shots Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models. Technically, Mainline RVC does have a go-realtime.bat (aka RVC-GUI), but it's pretty messy and outdated so it's extremely not suggested for realtime.

Wokada = uses RVC for realtime inference. There's 2 main versions, Original made by Wok, and the most suggested one is Deiteris Fork (modified version)

I'm guessing what you need is RVC, right?

[low shard]

you sure it's integrated graphics? if so, you're cooked, you will have to resort to cloud (remote good pc) with limited free time

AI is complex and intensive

[mellow notch]

If its realtime its better can i run wokada?

[mellow notch]

oh arent there any other live voice changers i could use ur models with on my pc locally

without cloud services

[low shard]

None, your pc is too weak, you can't expect AI to run on such weak hardware, either cloud or get a better pc

voicemod only gives 3 random non ai effects like robot or etc, voice.ai abuses your pc in background 24/7 and makes you pay too

you can't do anything else than either use cloud or buy a better pc for any ai tasks

[mellow notch]

voicemod is ai powered? and can i run ur models on voicemod?

[low shard]

what do you want to do exactly? you said about an anime serie, which needs inference (use models) on pre-recorded audios, not realtime

can you elaborate exactly?

[low shard]

nope, it just gives 3 random effects on free tier

trust me if there was a better alternative we would be suggesting it lol

[mellow notch]

main aim of finding a voice changer is that im still working on it (will take 2-3 more weeks) i just want to talk to my freinds and tryout different models till then

[storm fern]

Could some one help me

[low shard]

are you planning to use it for e girl trolling/catfish or just play with friends?

[low shard]

Sure,

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[mellow notch]

i dont play games i just wanna talk and try models with my school freinds

[low shard]

Alright, great

Your pc is too Weak for Wokada locally, You got 3 options:

• Buy a better pc
• Run it locally (on ur pc) using the CPU mode of the wokada fork which has better performance https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ (but this isn't suggested as it could be unstable)
• 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) :
  • W-Okada's Deiteris' Fork Voice Changer Google Colab (currently works only on google colab PAID tier)
  • How to use Original W-Okada's Voice Changer Google Colab (has a Guide) (currently broken)
• Kaggles (30 hours weekly of better GPUs, T4x2 & P100, harder to use, requires an account and a phone number):
  • W-Okada's Deiteris' Fork Voice Changer Kaggle (the best and only working one currently for free)
  • Original W-Okada's Voice Changer Kaggle (currently broken)

AI is complex and intensive, either just use cloud with limited free time or buy a better pc

I mean, 30 hours are good enough weekly, aren't they?

[storm fern]

I need help with the virtual audio cable things like I installed them but it won’t work and I don’t know what to do since the tutorial website you guys provide only provided a brief explanation pls and thank you

[mellow notch]

yeah can u tell me bout the cpu mode of wokanda i can run locally

[low shard]

that one is extremely unstable, you could have overheating issues and lagging asf, it's the least suggested option, if you want, you can try it by reading the link I already sent, but I already warn you

CPU isn't meant for running such complex tasks

your cpu isn't the worst like one from 2009, but still ehh I would really NOT suggest it, it isn't as good as having a dedicated gpu

Why don't you just try cloud? aren't 30 hours enough in a week?

[storm fern]

Pls and thank you

[low shard]

Hello, please elaborate what I asked, I need to know those info to help you and to be sure you're using the right guide and your pc is good enough

[storm fern]

I have a i7 with a 4060 ti also do you mean like operative system like you pc system cause am using windows 11 and https://youtu.be/LX5en3pZJwM?si=F1C4QImfYy-OT-62 and for the virtual audio cables am using you guys guide

[low shard]

that isn't our tutorial lol

that's an outdated video

there's no updated video for realtime voice changers

[flint vapor]

Yes , but you have to use a new updated version, as deiteris fork, that is a old outdated version of the voice changer

[low shard]

and there will never be

[flint vapor]

-wokada

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[flint vapor]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ Follow this guide

And download the one for nvidia series

NOT 5000

[low shard]

also, beatrice v2 models have lower quality

[storm fern]

Can you maybe provide me with one lol

[low shard]

May I ask what you want to do? like e girl trolling/catfishing, ai covers or roleplay?
to be sure to give you the right program

[storm fern]

Am using it for like trolling my hb or trolling on Roblox vc like using sponge bob or ghost face that kinda stuff

[low shard]

okay so no e girl trolling/catfishing in games/vc, right?

[storm fern]

No

[low shard]

great then, since catfishing is bad

[low shard]

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[storm fern]

Yea that’s weird

[low shard]

1st link, wokada deiteris fork

[low shard]

which btw is the same link given by the other member above 😭

[flint vapor]

There is the guide we send 😄 install the b2332

[hasty kite]

hi, what is the best option right now to train your own models?

[low shard]

This is a General AI Server, AI has many fields

Elaborate:

• your PC GPU
• your operative system
• what you want to do

[hasty kite]

oops, i have an rx 6600, win 11, i want to train my own voice model for tts/singing

[low shard]

tts/singing
Those require 2 different technology,
Text To Speech
&
Speech To Speech

I'm guessing you want to train RVC models right? RVC is only Speech To Speech natively, the only way to use it for TTS is via using another tts program as an input and then applying rvc on top, which isn't the best for emotions, but it's automated by an RVC Fork like Applio

Are you looking to do that?

[storm fern]

thx

[hasty kite]

yup!

[low shard]

check https://docs.applio.org/applio/getting-started/installation#amd-gpu-support-windows

[hasty kite]

thank you so much!

[storm fern]

ok so i got installed and what should i put as the input , output, and moniter?

[low shard]

you're welcome, be aware that it is more complex because AMD doesn't have great AI support btw

[storm fern]

cause to be real idk how