✨│ai-help

[covert valve]

i did all what u said

the AI guy sent me a link and i downloaded it

[low shard]

there's no updated video because ai moves at sonic speed

[low shard]

@flint vapor is not ai 😭

[low shard]

he's just an ex staff

[covert valve]

he heartbroken?

nick im sorry for being dumb

[flint vapor]

Why i'm an ai guy? haha

[low shard]

" 💔 " is used as a way to laugh sometimes in modern slang english

[covert valve]

oh

im getting hacked or what mr AI

this is the file link you sent me

[flint vapor]

Is just installing

[low shard]

it's just installing internal files

how good is your wifi connection?

[viral mason]

Thank you 🙏

[simple ore]

I swear, the ipad generation can't even read

[viral mason]

They can't

I'm not even Ipad generation but I need a video tutorial on some stuff just because I can't comprehend some stuff

[uncut haven]

https://i.imgur.com/PVpxwJb.png

[low shard]

don't click the 1st link

check the virtual audio cable step

then the download on nvidia windows one

!give-media-perms 1h @uncut haven

[uncut haven]

I'm downloading it again from this website

[low shard]

no need to download that

[uncut haven]

o.o

What then

[low shard]

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

[uncut haven]

I'm doing this rn

[low shard]

alr

[manic aspen]

wanting to make a chat bot AI that can past the turing test BUT also be unique

any ideas 🧐

[warm cedar]

i’m just curious what the difference between the index and pth whatever file is

[low shard]

Open the folder, clock the .exe

Please read the guide

[uncut haven]

My brain is too dumb for them

[low shard]

in rvc context:

• pth files: contain the voice
• added index files: contain the accent
• metadata.sjon file: it's just some extra info about the model download link if you downloaded it off weights.com, it's not needed and won't impact the actual model at all

[low shard]

Open Source AI is intensive and complex, don't expect it to work as an easy program like ChatGPT, it isn't made by a company to be a product 😭

[warm cedar]

thank you i was really curious

[low shard]

You can read it and tell me for any parts you don't understand ofc

[low shard]

Yw

[uncut haven]

Can I lower the ms without losing quality?

[simple ore]

if only people read the guides...

yes you can, as long as the meter stays green

if you wanna play games and use vc at the same time, then 128 ms chunk + 2.7s extra

or less if you dont

[low shard]

Uninstall vb audio cable it causes issues on windows
Get vac lite from the guide

Use rmvpe without onnx as far
Set extra to 2.7, chunk has to be higher than perf value

@uncut haven on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Reduce the delay via: https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

Also, read all the guide to understand how the program works please

[uncut haven]

I'm just a lazy person and don't like to read :3

No hate

But have respect

[wet obsidian]

can u help me pls

[wet obsidian]

So basically I've seen this problem more than often i need help with something

[low shard]

no need to ping me twice, if im busy i wont reply btw

now im back

https://discord.com/channels/1159260121998827560/1392912997151670272

use the help forum

im not sure why that other person told you to not use it

its right that u use the help post

[languid fossil]

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[wet obsidian]

uhm can somebody help with delay>>>

[lucid frigate]

guys, does anyone know a way to clone celbrities' voices in russian?

[flint vapor]

There is not a language option, you can clone in every language you want, you not cloning the language but the voice 😄

[lucid frigate]

I'm sorry for expressing myself wrong way (I'm not a native English speaker) I tried to say that I need to clone some celebrity's voice but there is no such a model in this server, and I need this cloning voice to be able to speak Russian

[flint vapor]

Every voice if you put then an audio as input that speak russian will speak russian

[flint vapor]

You asking help on how to clone a voice on RVC?

[lucid frigate]

yeah

[flint vapor]

Wait i will send you a guide and explain you 😄

[flint vapor]

Wait before send you guide, which GPU do you have?

[lucid frigate]

rtx 3050 4GB (laptop)

[flint vapor]

Unfortunately for train models in local it will not be so good, i will send you guide on how to do on cloud

[flint vapor]

https://docs.aihub.gg/rvc/cloud/applio-colab/ i don't know if this is the correct version to use in 2025 but you have to excuse me because i always trained in local and is not my field so much

But this is a guide on how run applio rvc in cloud

There will be some limits of usage

[flint vapor]

I excuse with moderators and staff if this isn' the correct version

[lucid frigate]

thanks a lot!

[flint vapor]

You welcome! 😄

[low shard]

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)

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]

Already helped in the help post btw

[flint vapor]

But there are also many, on the guide i sent you there are various for cloud section

[languid fossil]

One message removed from a suspended account.

[low shard]

That one version is over a year old original wokada

And vb audio cable causes issues on windows

don't trust ever video tutorials for realtime voice changers, they are all old

[languid fossil]

One message removed from a suspended account.

[low shard]

Whenever you want one or upgrade yours, always check there instead, ai is too fast for video tutorials

[languid fossil]

One message removed from a suspended account.

One message removed from a suspended account.

[low shard]

Every voice changer tutorial on YouTube is outdated

[low shard]

Also, that GPU is way too old

[languid fossil]

One message removed from a suspended account.

[low shard]

That's not good

Open source ai is a complex and intensive task

[languid fossil]

One message removed from a suspended account.

[low shard]

This runs on your hardware

It's made by the community

It's not a site running on a cloud server by a big ass company like chatgpt

It's free and open source, so it runs on your pc

[languid fossil]

One message removed from a suspended account.

[low shard]

Yes

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)

Also you got barely 2gb VRAM, 8gb VRAM isn't even enough for today

That GPU is like 13 years old

[languid fossil]

One message removed from a suspended account.

[low shard]

What's your CPU?

[languid fossil]

One message removed from a suspended account.

[low shard]

Ai is expensive, it can't run on low hardware

Try the options I said above

[low shard]

Damn that's old ASF too, 2nd generation

Your only options are either buy a better PC or use cloud which is more complex with limited free time

Don't even waste time trying to use your CPU or GPU, they are extremely old, literally more than a decade

It would be just wasted time to try locally

[languid fossil]

One message removed from a suspended account.

[low shard]

Alright let me know, are you going to buy a better PC or use cloud?

[languid fossil]

One message removed from a suspended account.

[low shard]

I mean cloud is free, you'd just need to read a guide, I think 30 hours weekly for free ain't bad

But your choice

[languid fossil]

One message removed from a suspended account.

[viral mason]

I could show u how to use applio kaggle in like 5 seconds

it ain't that hard since u can't run it locally

[low shard]

We aren't talking about applio

😭

It was about wokada deiteris fork

Anyways his choice

[viral mason]

o

I'm tired as hell rn bc I'm being made to work out in the heat rn and just got back so idk what's goin on

[warm cedar]

can i delete the models i put onto w okada and still have them on w okada

like if i deleted them from my files (the index and pth) would they still be usable on the thing?

[low shard]

you mean the model files in the folder "model_dir"? no they wouldn't still be usable

[warm cedar]

no i mean the folder you download originally and the extracted folder after that

like if i downloaded a mrbeast voice changer it would be a folder, then i would extract that folder, then i’d put the index and pth from said folder onto w okada; so my question is can i get rid of the first two folders from the start of the process

[low shard]

mrbeast voice changer
I think you mean RVC model

and yes, you could delete them after uploading them into w-okada deiteris fork, because they will be copied into an internal folder called "model_dir"

btw be sure you're using wokada deiteris fork and not original wokada

[warm cedar]

idk what the difference is

is there a problem with the original? i think i downloaded it off of face hugger

something like that

hold on i’ll find the video i used to download it

https://youtu.be/VAtnHB9Y9UA?si=7eyZnZeUqoyyREo0

[hazy eagle]

i have question again, should i turn off game mode?

does the forkVC detects as background apps?

[low shard]

holy shitt, that's extremely outdated, the video uses an over year old version of origianl wokada, and vb audio cable which causes issues on windows

[low shard]

literally everything, wokada deiteris fork has better quality and performance, and isn't as buggy

plus the settings suggested in the videos are extremely bad 😭

please just forget everything you saw in that video, uninstall everything

harvest isn't used from like 2023 😭

[warm cedar]

aw man do you have a tutorial you can send me on how to download it? i also did download the vb cables used :(

[low shard]

could you please tell me your pc gpu? i'm guessing you're on windows

im just asking so i can give you the updated link

[warm cedar]

4060? is that the gpu or cpu idk

[low shard]

nvidia geforce rtx 4060? yep that's a GPU, a good one

uninstall everything

[warm cedar]

yeah

[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, read up wokada deiteris fork

[warm cedar]

oh i really can’t figure out how to do it without a video lol sorry i’m just not smart like that

in a zombie apocalypse i would not survive

[low shard]

you shouldn't trust ever video tutorials for realtime voice changers, they always end up being outdated, while written guides can get updated easily

[low shard]

wait do you mean you have like Dyslexia ?

[warm cedar]

lol nah i do have adhd though (diagnosed) and it just makes it really hard to contextualize words i read

[low shard]

I could help you set it up via text if you want, tho I can't VC

[warm cedar]

it’s like talking at a brick wall

okay

gimme a sec to hop on my pc

[low shard]

alr lmk

first of all uninstall everything

then download vac lite (which is better than vb audio cable): https://software.muzychenko.net/freeware/vac470lite.zip

[warm cedar]

okay it's installed and i already extracted the files

[hazy eagle]

anyone?

[warm cedar]

and i've deleted the original program and the zip and the old vb

[warm cedar]

oh wait i deleted the vb but it still shows up as an option

what should i do?

[flint vapor]

You went in device manager for uninstall it?

[warm cedar]

now i have

thyank you lol

[flint vapor]

Now after you uninstall vb you need vac lite 😁

https://software.muzychenko.net/freeware/vac470lite.zip

[warm cedar]

oh yeah i have it installed and unzipped but what do i do with it now

[flint vapor]

Run the setup64

[warm cedar]

64a or 64

[flint vapor]

64

[warm cedar]

kk it'

it's installing

[hazy eagle]

[low shard]

yes

[warm cedar]

ok it's installed

[low shard]

did you do the system audio properties step?

it should have opened a window after clicking yes

[warm cedar]

can i delete the file folder and compressed folder

[warm cedar]

it did but i just clicked okay was i supposed to do something

[low shard]

yep you were lol

did you close it already?

[warm cedar]

yeah

[low shard]

if so, you have to reopen it via:
windows+r
type "mmsys.cpl" then enter

[warm cedar]

okay, now what?

[low shard]

then go in the recording tab, set your usual input device as default, and right click it and make it default communications devices too

do the same thing to your usual output device in the playback tab

[warm cedar]

if you're wondering why i use voicemod my mic is quiet

[low shard]

wasn't voicemod just 3 like daily random free special effects?

[warm cedar]

it was but i have lifetime premium and it was only $20 cause it was a deal for switching from v2 to v3

[low shard]

btw using voicemod could technically cause more delay, and you could already change your volume audio options in the wokada deiteris fork, but your choice

[warm cedar]

anyways i right clicked it and i dont see default communications device

[edgy minnow]

tyyy

[warm cedar]

what do i do now?

[low shard]

did you do the same for the playback tab?

[warm cedar]

yeah

[low shard]

click ok

[warm cedar]

kk

[low shard]

download https://huggingface.co/Shadicti/deiteris-Fork/blob/main/voice-changer-windows-nvidia-b2332.zip

[warm cedar]

it's downloading

while its doing that can i ask if the thing i was using was really that old lol

okay it's downloaded, unzip?

[flint vapor]

Yes😁

[warm cedar]

im honestly nervous to do anything without confirmation cause i mess stuff up

[flint vapor]

Don’t worry you will not mess up

Now after you have unzip

Click on mmvc folder

And run the exe

[low shard]

yeah 😭

[flint vapor]

It will take some seconds to open and it will download some models don’t be scared if you see some things in terminal

[warm cedar]

ok i kept all my old rv so i could load em on they should work right?

[flint vapor]

Yes If it stored your pth and index you can upload then

[flint vapor]

Ok now open the folder 😁

And run the exe

[warm cedar]

yhe top one?

[flint vapor]

Wait some seconds the terminal will open and it will download some models for work

Yes 😁

[low shard]

@flint vapor btw you wanted to become an helper again? I see recently you're trying to help after a long time I haven't seen you lol

[warm cedar]

it's open but why is it in a browser and not it's own application

[low shard]

it just uses a web user interface like most ai programs, it's normal don't worry, it still runs locally

there's a part that goes more in detail about the reason in https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window but idk if you want to read it, so i gave you a short response lol

[low shard]

btw what browser are u using?

[flint vapor]

Oh in this period I been more free and I was interested in AI again and I wanted give some help In chats how I was did 1 year ago hahah

[warm cedar]

it's the norton one i uninstalled norton but the browser is still here i guess...

[low shard]

input: microphone
output: line 1
monitor: optional to headphones to hear urself
gpu: rtx 4060
extra: 2.7
chunk: 90ms
f0: rmvpe without onnx

[low shard]

some anti viruses are worse than viruses

also im not sure if that could cause some issues with wokada

[flint vapor]

It’s like, in a way, I have started wanting AI again.

[warm cedar]

is there an easy way to switch it to like microsoft edge or should i not bother

[flint vapor]

I recommend to you remove all about Norton

Because is considered a malware nowadays

[warm cedar]

i have this is the last remnance because norton is like an infected leg you gotta chop it off

[low shard]

you can just copy the url to another browser, but it would be the best to uninstall it https://support.norton.com/sp/en/us/home/current/solutions/v15972972

[low shard]

they had to literally make a tool to uninstall their own program 💀

[flint vapor]

Yes Norton is really a malware, and even if you remove it you have to remove manually all his tracks

Before also Kaspersky was say okada was a malware 2 years ago but then they updated it I think and it was not anymore flagged as a malware

[warm cedar]

ive already done all that its just the browser

[low shard]

Ohh i see

wait do you have a job?

[low shard]

maybe try in the windows app settings 😭

im shocked lmao

[warm cedar]

revo uninstaller my beloved

[flint vapor]

Oh no no still always at school, in Italy school finish even at 19 😭 but now I free, is not only about school but I was lost a bit the interest before but then I saw videos about flux kontext, veo, various updates, perplexity, and I wanted again use AI stuff, I remember I was in this server and know info about rvc again haha

And I wanted to help again as I was do a year ago

Because I was really like but in the last period I was a bit lost the interest

[low shard]

🙏

[low shard]

i thought u were older for a sec

did u forget im italian btw 😭

[flint vapor]

No no hahah don’t worry, I remember when we were talking in the Italian chat 2 years ago

[low shard]

btw we are trying to expand from being just an rvc server recently

which is why we have unified ai help chats

[flint vapor]

Is very nostalgic sometime think to that moments

[flint vapor]

Yes I saw now is all in English and the Italian server died 😭

[low shard]

the italian chat got reoved for being dead asf

[flint vapor]

Yes 😭in 2023 I remember it was so full I think it was one of the most active I think you remember too haha

[warm cedar]

guys what settings should i use

[flint vapor]

For first remove cpu and select your gpu

[low shard]

#✨│ai-help message

[flint vapor]

@warm cedar

[low shard]

yep lol

[warm cedar]

bro we're in that chat rn

😭

[flint vapor]

I remember with Ilaria when she was Admin

However I saw she not have the role contributor 😭

[warm cedar]

nvm i just thought you sent this channel

anyways thank you guys so much!!

i hope it doesnt lag out my fortnite

[flint vapor]

If you can add role you can put her if you want, I noticed she only have model maker

[low shard]

she didn't want it iirc

[flint vapor]

Ohh I didn’t know this sorry

[low shard]

its fine dw

[low shard]

yeah a lot of things changed btw

[flint vapor]

Yes I saw is not focused only on voices anymore now but ai in general

[low shard]

especially staff

[flint vapor]

Yes I saw some people that were before staff they are neither in the server anymore

[low shard]

yeah bc RVC kinda died, the og devs left it to rot to go work on another TTS program instead

[flint vapor]

Yes if I not wrong they working now on gpt so vits

That is also very good I tried but it just have 4 languages

[low shard]

yup that exactly lol

[flint vapor]

Yes is so sad, we still stuck at v2 after 2 years, RVC was a really good idea I hope someday they will work on again

[low shard]

i mean there are some rvc forks and experiments, but its not rvc v3

whats your opinion on the server expanding btw?

[warm cedar]

im worried its running worse than the old program... can you guys confirm these settings are right

[flint vapor]

Oh I’m very happy is expanding also to other type of AI and it looks also more organized than before, unfortunately the Italian community died but the server is very going great and I’m happy is expanding well 😁

[low shard]

those settings seem fine, could you explain why do you say that? you don't seem lagging, tho you missed the optional better settings

on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability/quality. Increases VRAM usage by 200 MB)
• Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Reduce the delay via: https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay (tho this is a bit more complex)

[warm cedar]

how do i do the first 2

[low shard]

I'm glad to hear that! when you feel like you want to be helper again and know about enough things, feel free to apply

[low shard]

you click on the advanced settings button, and the settings are literally there lol

[flint vapor]

Thank you so much! 😁 Yes if I will be active also these days I would for sure try to apply again

[low shard]

force fp32 mode: on
disable jit compilation: on

[warm cedar]

they're both grey i can't click em

[low shard]

oops, close, then click stop, then go back to that screen

[warm cedar]

ohhhh

it seems more crackly now maybe my computer can't handle it

[low shard]

could you show a screenshot of ur wokada?

does it crack always, or only on like discord?

[low shard]

seems fine, does it always crack, or only when used on discord ?

does it actually crack, or just has some cutoff issues ?

[warm cedar]

im not using it on discord or on call or anything i hear it through the monitor

it crackles and also cuts off

[low shard]

set extra to 2.0

then for cracking:
Open Task Manager > Details

Right click audiodg.exe and set priority to High

Right click audiodg.exe again > set affinity > uncheck everything except CPU 2 (only ✅ CPU 2, turn off the rest)

[warm cedar]

really? only keep cpu 2? all 23?

just making sure

yo it's so much worse now LOL

[low shard]

are you sure you had crackles, or just cuts off 😭

[warm cedar]

100% it crackles and also cuts off

it's popping

like bubble wrap

or it kinda sounds like that noise when your science teacher hits 2 magnets together to show you how magnets work you know?

you get em from chuck e cheese for 200 tokens

[low shard]

have you also tried with other models?

[warm cedar]

no

[low shard]

try other models

not all models are good quality

[warm cedar]

yeah i tried another model and it's just as bad, when i used them on the other program they didn't glitch as well

[low shard]

wtf? can u show ur full settings again

[warm cedar]

i restarted it and now it's working lol

[low shard]

now it's all fine?

[warm cedar]

i just re-intialized it

yeah

[low shard]

turning off and on fixes everythingg ong 🔥

[warm cedar]

thanks bro

[low shard]

yeah sounds better now

[low shard]

you're welcome, need any other help?

[warm cedar]

nah all good everything else is easy

[low shard]

Alright, be sure you also set line 1 as the input in discord or whatever other program you're gonna use

[warm cedar]

i made that mistake last night with the other program and i was using my normal microphone when trying to be peter griffon, and you know with voice like peter griffin you have to try to do an impression for it to sound good and i kept saying stuff and people were acting like i was weird and i didn’t understand till i checked my input 😭

god so embarrassing

luckily it was online so it’s whatever

[low shard]

LMAO 😭

[low shard]

is this u on his video lmao

also i wonder what happened to fe4rless lol

[warm cedar]

this is annoying me

i can't change model names or delete them

[low shard]

about renaming, you have to rename the model file, and then re-upload it in the slot

for deleting, you have to either overwrite the slot, or delete it from the folder "model_dir"

[warm cedar]

thanks

[warm cedar]

yep

[low shard]

you're welcome

[warm cedar]

what voices do u think i should use?

[viral mason]

freddy fazbear

[manic aspen]

Any good app/place i can program my first chat bot

?

[low shard]

there's thousands of voices, whatever you want like markiplier lol

[broken arrow]

im using rvc and some models for some reason are just really low volume even if i max out the output, what should i do?

[uncut cairn]

can someone help me im new

[simple ore]

need to install torch cu128 in whatever application you're trying to use

[viral mason]

I have that issue too

[pastel bison]

okada isnt working on discord

ive put the okada input as my mic, the output as cable input. ive also put the discord input as cable output and the output as my headphones

[abstract rain]

okada delay and audio cracks despite decent spec (amd gpu tho)

[heady sinew]

can somebody help me my voice changer is not working

[civic anvil]

how i uninstall this voicechanger

[civic anvil]

?

[latent kettle]

Delete the folder

[brittle wing]

What settings should I use on Audio Sr to upscale lq recordings to sound more cdq

[civic anvil]

is it uninstall from my systum after delete folder ?

[latent kettle]

Everything is included in the folder

[flint vapor]

Yes just delete all folder

And then empty the cna

*can

[latent kettle]

So if you delete that folder it will be deleted

[hazy eagle]

should i off my game mode or nah? for my forkVC

[flint vapor]

What error do you have?

[hazy eagle]

does game mode detects forkVC as Background apps?

[flint vapor]

However is from a lot i don't use voice changer, i downloaded now deiteris is all fine but the problem is that when i do start i don't hear the audio in client. my input is my default mic, as output i have put line 1, monitor i put my headphones, i want only do test for listen my self. but i can hear my voice just in server, when i select client not, with the old wokada with same audio configuration was working, but maybe i wrong something idk

i have also tried put on fp32 and disable jit

Bascially vol(in) is to -90 db while in server is not stuck

Oh now i think start working again, i think maybe it was a little bug

[gloomy lynx]

Probably a really stupid question, but how likely are you to trigger implicit/explicit caching when using Gemini's API (via SillyTavern) for creative writing? Wanna make sure whether there's anything I can do on my end to save a buck or two

[simple ore]

you can ask this question in silly tavern discord 😛

[hollow crater]

Hey guys , who is using ai for content creation here?

[brittle wing]

Hello, when I use the RVC GUI to set GPU (RTX 5060), program closes and doesn't convert, but when I use the CPU, it does. Is there a solution to this? I think the RTX 5000 series is having problems.

[junior agate]

HOW DO I DOWNLOAD VC??

[hallow thistle]

Most of RVC programs as of today won't gonna work with NVIDIA GeForce RTX 50 series GPU. Applio RVC is the only known RVC that can work with RTX 50, but with more steps than the regular one. See this message for guide. #✨│ai-help message

[hallow thistle]

!howtoask

[patent trellis]

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.

[hallow thistle]

What is your PC GPU? Yes, there's a way to download either RVC or W-Okada, but it won't be what from tutorial videos on YouTube.

[brittle wing]

Thanks, I'll check it out soon ^^. I had a similar problem with w-okada, but I solved it. It's really annoying to have the latest technology and deal with things like this :(

[hallow thistle]

There's one and only W-Okada version that is made for RTX 50 especially. You can check this out if you haven't yet. https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#download-nvidia-rtx-5000-series-on-windows

[junior agate]

what's the latest version of VC?

[hallow thistle]

Deiteris fork W-Okada b2332 is the latest version. The word "VC" can either mean "voice chat" and "voice changer".

Anyways, what is your PC GPU again?

[knotty moth]

-gui

[patent trellis]

RVC Easy GUI

This program is pretty abandoned and old, don't use it.

[knotty moth]

esp RTX 50-series gpus don't have backward compatibility to any AI applications not updated to use pytorch 2.7 and cuda 12.8

if you still want to use that old application, why not just run in cpu mode

[flint vapor]

Is this tool still exist?😭

-svc

[patent trellis]

https://media.discordapp.net/attachments/1159822429137412177/1194008385113301003/svc.gif?ex=67f2ddf6&is=67f18c76&hm=17bc260b884736770ef6374560c253ce24b348309d2ee35530100af590aef719&=&width=614&height=561

[brittle wing]

thanks i tried this yesterday and it worked and now i will try the thing on rvc gui

[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

reminder that there's no updated video tutorial on youtube

[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]

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)

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

I don't really understand what you're talking about

[low shard]

RVC GUI is old asf, all youtube tuts are outdated for rvc

the other guy already gave you the link

[low shard]

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)

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]

rvc gui is outdated asf, don't even bother using it bro

just use applio

rvc gui is abandoned since more than 2 years now

[hallow thistle]

"RVC GUI" refers to varity of RVC programs that are developed to feature GUI as its main interface. which includes Applio, RVC Easy GUI and even Mainline RVC. It doesn't refer to a specific program. But if you mean by an old program like from 2023, as what I said, it still won't gonna work with RTX 50 GPU anyway.

Any RVC program will always work with any GTX 10, RTX 20, 30 and 40 GPUs. RTX 50 is too new, so don't expect every AI program to work like that. It would be better to stick with Applio on this one.

[fleet cedar]

what happens if i put extra 5s

is that recommended

[low shard]

extra controls a bit the quality, if you put it over 2.7, it can cause cutoff issues, and ofc takes more performance

[fleet cedar]

dont see any issues or delay tbh

[low shard]

its not recommended to put it over 2.7, because of the cutoff issues

like at the end of your words

tho those might depend on some models

its usually the best to just keep it at 2.7

[fleet cedar]

alr, its on 2.7 now by the way my accent is pretty strong

is it fine if i put it to index 1

or smth

[low shard]

index controls the accent, the higher the value the more its closer to the trained accent, tho be aware that can also give some autotune effect

do you want to show a screenshot of ur whole wokada

so i can check ur entire settings

[low shard]

why is the out and mon so loud 😭

[fleet cedar]

cause ts is so quiet

depending on the voice model i change it to lower

[low shard]

mon is just how you hear urself btw

pitch ofc depends on ur voice and model

ofcourse its suggested to use vac lite instead of vb audio cable for crackle and randomly stop working issues reported by many users

on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Reduce the delay via: https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

[fleet cedar]

ik, btw vb audio cable works more smoothly for me than vac lite

[low shard]

you're the first that says that in 2 years, are you really sure you just didn't fuck up the vac lite installation?

[fleet cedar]

i am sure

i followed the guide

exactly

[brittle wing]

I understand, I understand, what I mean is I fixed w-okada, now it's time to fix this.

[low shard]

you do you then, just know that it can stop randomly working

[low shard]

do you need help to get wokada deiteris fork?

[fleet cedar]

if it does, ill try vac lite and fix it

[low shard]

anyways try the optional benefits

[fleet cedar]

okay

[brittle wing]

Thanks, but luckily this (https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#download-nvidia-rtx-5000-series-on-windows) worked without any problems, I can use my models comfortably. <3

[low shard]

alright, if there's ever any issues let us know here

!howtoask

[patent trellis]

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.

[brittle wing]

[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

[ivory venture]

guys im trying to make a knowldege based on previous exams to remake new ones is this what i really need?
Tools for model training and orchestration: LlamaIndex or langchain
Tools for model training and orchestration: pinecorn?

[quasi rover]

hi yall, 9 minutes dataset of clean rapper vocals is good for a good model?

also, how many epochs?

[low shard]

there's thousands of programs I could call simply voice changer

elaborate all I asked

there's also hundreds of amd gpus, you need to elaborate which one

[quasi rover]

please, help.

[low shard]

you could be using a modern one or an ancient one

[low shard]

no one knows, there's no right epochs amount https://docs.aihub.gg/rvc/resources/training/#epochs--overtraining

but yeah 9 minutes is not bad

[quasi rover]

alright thanks, cause 3 years ago I previously trained a model under 5 minutes of dataset and it came out high quality, guess its quality over quantity. I tried to use other vocals but one song was great for that to go well

[low shard]

that tutorial is outdated asf

check the comments next time

there's no updated video tutorial for voice changers lol

unfortunately yes

you can uninstall everything you got off that tutorial, all outdated

-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

is way better in quality and performance

[ivory venture]

no ideas?

[low shard]

forget everything you saw on that tutorial

[hallow thistle]

Yes, I am a human and a real person. If you see me replying fast, it's because I type somewhat fast.

[ivory venture]

the reply kinda unhuman

[quasi rover]

they are AIs bro why u asking this in an AI discord

[low shard]

sorry but you might need to wait for an helper that knows more on that subject, might be better to ask @simple ore or post in in #1192011222023950368

[hallow thistle]

[knotty moth]

it still works but longer dataset with more variation while keeping consistent may be better

[low shard]

we're not AI gng 💔

[hallow thistle]

Well, I see that's an issue, I guess.

[quasi rover]

damn AI can use brainrot words is crazy

[ivory venture]

can i go on his DM or just wait him?But got u

[low shard]

uninstall both that old original wokada and both vb audio cable

then read the guide and let me know

[quasi rover]

thank you bro I'll try to get more songs. I used one 10 minutes song but it came out 9 minutes coz I removed silence and shit

[low shard]

maybe just wait

[quasi rover]

luckily the artist is a storyteller so his songs are 5+ minutes

[hallow thistle]

If you see the word "AI" next to my name, it's a "server tag" of AI Hub by Weights.

[low shard]

-# he's clearly AI trust me 🔥

just delete the zip and folder of wokada
then uninstall vb audio cable from the windows app settings

[hallow thistle]

While I'm aware the "AI chatbot" accusation is sarcastic, I would still answer in a serious matter anyway.

[quasi rover]

do you know any vocal splitter for free? I used all my free gpu for colab and huggingface

I aint tryna use UVR cause my gpu is shit

[low shard]

what's ur pc gpu?

[quasi rover]

help me find it, i just know is shit

can't handle gta san andreas bro

[low shard]

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

you might have 2 gpus

[low shard]

... what?

you must be joking right

that must be sarcastic

[quasi rover]

nah bro I play lsrp on samp and that shit is so laggy

[low shard]

you can't possibly be running on a early 2000s pc

[quasi rover]

dont even know why it is like that

[quasi rover]

loool

well let me see my gpu ill tell you inabit

[low shard]

what windows version are you on? use the method i told u to check ur pc gpu

yes the wokada deiteris fork is better for amd

[quasi rover]

well it just says

AMD Radeon (TM) Graphics

[low shard]

that's integrated graphics, do you have any other GPU?

[quasi rover]

nope.

[quasi rover]

I honestly bought this PC for making samples and beats 😦

[low shard]

well your pc is cooked

you can't expect everything for free, https://docs.aihub.gg/rvc/resources/dataset-isolation/#vocal-isolation--cleaning if you finished the free gpu of every single cloud thing mentioned here, either wait or pay for colab

[quasi rover]

I had a better PC before where I could train models with my gpu but i sold it

[quasi rover]

Loooooool

[low shard]

wdym? show a screenshot

[quasi rover]

told you bro i cant handle gta sa

[low shard]

!give-media-perms 1h @strong gyro

[low shard]

just upgrade atp

[quasi rover]

yeah I remember there was a website

[low shard]

ofc cloud can't give you everything unlimited for free

[quasi rover]

for gpu

8 € per month

something like that

[low shard]

no you shouldn't click the github link.. that's just for telling the project repository.. you have to read the guide

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

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#download-amd-intel-and-cpu-on-windows

read it carefully, else it can fuck up your whole audio system

[hallow thistle]

"AMD Radeon" but without "RX" in its name is an integrated GPU. This integrated AMD GPU always come with AMD Ryzen CPU, much like Intel UHD Graphics which always comes with an Intel CPU.

[low shard]

https://colab.research.google.com/signup

but u should really upgrade ur pc instead

also its not 8 euros per month, its more expensive than that

[hallow thistle]

Any integrated GPU is not suitable for AI. If your PC only has an integrated GPU, there's no choice other than use a cloud service anyway.

[brittle wing]

also Thanks for your interest. AI Voice is really fun and cool, I plan to train my own models soon. I think I've downloaded everything I need for that (w-okada - rvc - vocal remover etc)

[low shard]

Alright you're welcome

setup64

[hallow thistle]

This fork W-Okada is red.

[low shard]

be sure to click yes later on, else that is the step that could be fucked up and not make you hear anything on your pc

[hallow thistle]

"setup64.exe"

[heady sinew]

can somebody help me my voice changer is not working when i talk it does nothing

[hallow thistle]

!howtoask

[patent trellis]

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.

[hallow thistle]

Which W-Okada version are you using? Are you using VB-Cable? And what is your PC GPU?

[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

[heady sinew]

my gpu is a intelX irisR Xe Graphics and i am using vb-cable and idk wich version im on

[low shard]

that's integrated graphics

it's bad asf, powerless

lemme guess, you use da youtube tutorial? all video tutorials are outdated

[heady sinew]

i think so cidk

idk

[low shard]

do you have any other gpu?

[heady sinew]

no

[low shard]

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

are you sure?

[heady sinew]

11th Gen Intel(R) Core(TM) i5-1135G7 @ 2.40GHz thats my gpu

[low shard]

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)

[low shard]

that's a cpu, not a gpu

you can uninstall everything you have

[heady sinew]

no i dont have other gpu then

[low shard]

your pc is too weak to run it

[low shard]

then either buy a better pc or use cloud with limited free time

AI is intensive and complex

[heady sinew]

ok il take the free

one

[low shard]

you mean cloud? this one then

you can uninstall vb audio cable and everything you got before

[heady sinew]

yeah

[low shard]

read up that guide I sent you, there's no updated video tutorial for voice changers

read it carefully, it can fuck up your whole audio system if you do it wrong

[hallow thistle]

Like Intel UHD Graphics, Intel Iris Xe Graphics is another integrated GPU that comes with certain Intel CPUs. Any integrated GPU, like this one, is still not suitable for AI, even W-Okada. VB-Cable causes random issues for Windows users when using with W-Okada.

[balmy wing]

Sorry to interrupt. Speaking of gpu, should i go with amd or nvidia?, i know nvidia has cuda dominance, but is amd any good now?.

[low shard]

no, that's because you didn't do the 2nd step in the picture i sent you..

i told you

if you dont read it carefully

you can fuck up your whole audio 😭

do the 2nd step in the picture

[low shard]

What are you planning to do? which amd and nvidia gpu?

if you're talking about any AI task, Nvidia is better

tho a decent nvidia gpu, not an old one

[balmy wing]

The new 9060xt or 5060 ti 16 gb versions.

[low shard]

the nvidia one would be better if you're planning to do AI

non nvidia gpus don't have as good AI support

[hallow thistle]

NVIDIA GeForce RTX GPUs are more suitable for AIs because they have CUDA cores. AMD Radeon RX GPUs are also good, they usually don't come with CUDA core, but there are some SDKs that allow CUDA-related tasks to run on AMD Radeon RX GPUs.

[low shard]

that's just for linux

you need to follow only the general steps, and the steps for amd on windows

you're on windows, not linux

do you know what's linux?

Linux is a family of open-source Unix-like operating systems based on the Linux kernel

[flint vapor]

Yes

[low shard]

perfectly normal

it runs on your pc hardware

just uses a web user interface

read https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window if u care about knowing more about this

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window

they are both programs running on your hardware

[flint vapor]

Which? The old okada?

[low shard]

the only difference is how the interface opens

[flint vapor]

It was same a webui but closed in a specific window

[low shard]

even the interface are made in the same programming language

[hallow thistle]

If you see fork W-Okada opens a browser, it's completely normal. The "original" W-Okada opens itself as a separate program.

[flint vapor]

Is like an app too but using your web browser only as interface

[low shard]

nope, the great majority of AI programs use a web user interface

this is open soruce ai by the community, not a product made by a company like chatgpt

[hallow thistle]

This sentence is confusing to read. So, Fork W-Okada (b2332) opens a browser when launching, which is completely normal for this specific W-Okada. While the "original" W-Okada version (v.1.5.3.18a) opens as a separate program.

[flint vapor]

Yes sorry, i wanted mean that the original w-okada version open itself as separate program but is a web interface too if i not wrong, i saw it was connect to an url

Or maybe i confused some other tool

[viral mason]

egirl..

[viral mason]

https://tenor.com/view/goth-girl-goth-kiss-black-man-kiss-gif-1659276405462406042

this u

[hallow thistle]

GPU: AMD Radeon RX 6600 (DirectML)
Chunk: around 70 - 100 ms
Extra is always be 2.7 s.

[viral mason]

it doesn't like u

[hallow thistle]

Nah, I'm for real telling you where to improve audio quality on W-Okada.

[flint vapor]

I saw it uses always a local server and connect to local host running a small window but if i not wrong is a webui

[viral mason]

don't listen to my silly business I'm just making fun bc egrill model

[flint vapor]

Of course it work on a separate exe but i think it connects to my localhost because if i run the same url it will open also on browser

But i don't know if this is correct

[low shard]

Are you planning to catfish as an e girl in vcs?

[hallow thistle]

127.0.0.1 is a localhost loopback IP address found in every PC system. The xxxxx that located after 128.0.0.1 is a port number (127.0.0.1:xxxxx). The default port number for fork W-Okada is 18888, which in turn is 127.0.0.1:18888 or localhost:18888. While you can type the word "localhost" following with a port number instead of 127.0.0.1 in URL, it will still function the same.

[flint vapor]

Oh yes i know but isn't the separate exe connecting always to webui?

[hallow thistle]

I'm only here to guide on how to install and use W-Okada, not to catch a predator.

[flint vapor]

Because i saw i can do also inspect

On the old okada

However i still have to uninstall it idk why i still keep it

😭

[flint vapor]

Oh yes the port for okada Is 18888, sorry maybe i misunderstood the question i thought you were say that the old w okada interface was only in a separate exe

[flint vapor]

And not using the port

Sorry if i have misunderstood

[hallow thistle]

Yes, the original W-Okada program uses a kind of Chromium HTML engine, hence you may see the Inspect Element window when you press F12.

[flint vapor]

It isn't basically always catfishing?

[flint vapor]

You should be careful to troll that people

[low shard]

catfishers get banned

[flint vapor]

Yes

If he do to that people

Is always bascially

A catfish

[low shard]

yeah he's banned, for severe cases they should report bad people, not try to make money via videos on catfishing on vc

[flint vapor]

Yes true if he catfish he will not improve the situation and maybe also enter in some risks because it is dangerous even to talk to that people

[abstract rain]

sorry I fixed it with the deiteris fork

[low shard]

do you want me to check your settings?

[hallow thistle]

Using W-Okada the realtime voice changer in offensive situations (like catfishing) is indeed considered bad. While I only expect people here to use W-Okada for fun, or troll or confuse people for funny, saying you wanna use W-Okada to catfishing someone (e.g., pretend to be a girl to get Nitro for free) will not only make things worse.

[flint vapor]

@livid surge Which error or problem do you have with client?

[livid surge]

whenever i use it in a game

the res thing gets really high and whatever i say doesnt work

[worthy marlin]

I just updated w okada which I had a completely different version because it was two years old, who can suggest me some settings for the 3060

[flint vapor]

If you have both opened the performance of client will reduce a bit, which gpu do you have? I recommend you to put mid low quality in the game

While you have vc opened

[livid surge]

uhm

nvidia 1050

[flint vapor]

Oh 1050 is not suggested for using it on Okada

[livid surge]

like

[flint vapor]

You have 2 gb vram?

[livid surge]

no

4 gb

[hallow thistle]

Which W-Okada? Is it b2332 or v.1.5.3.18a? These are two different versions of W-Okada.

[livid surge]

like the entire thing works when i dont use it in a game

[flint vapor]

Oh is the same not suggested so much, it will be very slow

Especially while you run game

[livid surge]

works fine in discord too

but like ingame

[latent kettle]

Both w-okada and games use GPU that's why

[flint vapor]

Oh yes it can be work fine but when there is a game opened it will not work so fine, 1050 is not so much recommended for Okada so in games it can do some problems

[latent kettle]

And 1050 is not capable of handling Both at same time

[livid surge]

so do i make the

[latent kettle]

@livid surge

[livid surge]

voice changer thing

use my cpu

[latent kettle]

That gonna very slow.

[hallow thistle]

While it's possible to run W-Okada with any NVIDIA GeForce GTX 10 GPU, it can be slow when running alongside a graphics-demanding game because of how old GTX 10 is.

[worthy marlin]

I upgraded from 1.5.3.18a cuda to 2.0.78 beta cuda windows

[flint vapor]

I don't recommend you to put cpu or it will explode all while the game use your gpu and okada your cpu

[latent kettle]

Poor output and voice crakes

[flint vapor]

Oh yes sorry i wanted say while gaming sorry

[livid surge]

so then

i cant use the voice changer thing ingame?

[analog obsidian]

no

[latent kettle]

You can use limited, cloud alternatives

Such as kaggle

[hallow thistle]

2.0.78 is the true latest version of the "original" W-Okada, but is still not recommended to use. There's a better one to try, I use fork W-Okada (b2332).

[livid surge]

the res thing is 400ms when i dont use it

[worthy marlin]

can you send me the link?

[livid surge]

what chunks

do i use

i speak and i hear nothing

[hallow thistle]

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

[livid surge]

im using

192 (512.0ms)

i cant go lower cus then i wont hear anything

[worthy marlin]

thanks

[livid surge]

and it takes a few second to hear what i said

[latent kettle]

https://docs.aihub.gg/rvc-voice-changer/cloud/w-okada-kaggle/#w-okada-kaggle

[worthy marlin]

For the settings I use the same ones as the old version without problems or has something changed?

[livid surge]

that looks complicated

[latent kettle]

If you read it carefully it's not that complicated

Don't judge it by looking at it 🥲

[livid surge]

too much work for a voice changer

[flint vapor]

It is not don't worry only follow correct the guide is very simple

[latent kettle]

That's what I said, it's not that complicated, if you get any error ask here

[viral mason]

Dr. Jr is your name based off the spongebob ai thingy that makes Plankton glitch out?

[worthy marlin]

However, I can't find the http start in the file. How do I start it

[flint vapor]

There is not http start

There is a .exe in the folder mmvc

[hallow thistle]

Fork and original W-Okada are different. The "Extra" number on the original one may be "2048", but on fork W-Okada it's now "0.5 s".

[hallow thistle]

If Deiteris' fork W-Okada folder looks like this, it's completely normal so don't worry. The actual program is named "MMVCServerSIO.exe". The original one has a lot of files within the "MMVCServerSIO" folder, which is a mess.

[latent kettle]

No. It's from iron man's Ai jarvais

[hallow thistle]

Settings on your fork W-Okada:
Chunk: around 60 - 90 ms
Extra is always be 2.7 s.
GPU: NVIDIA GeForce RTX 3060
https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#finding-my-own-settings-for-chunk

[young halo]

does someonw know about how much does the applio colab runtime last?

[hallow thistle]

On Google Colab with free tier, running a notebook with T4 GPU usually lasts 4 hours average.

[young halo]

oh, it should be enough for the model i'm training then, thank u!

[viral mason]

just saying kaggle can run for 30 hours free (applio)

[young halo]

do u have any tutorial and link?

[viral mason]

I could help u in dms ^^

the link tho is here kaggle.com/code/deiant/applio

[hallow thistle]

Kaggle notebook with T4 x 2 GPUs usually lasts 30 hours a week.

[young halo]

i hope i can get it to at least 200 epochs before the connection breaks, cause i'm not sure abou how to resume it

[hallow thistle]

Google Colab Pro tier or "Pay As You Go" option, however, the runtime for each GPU (T4, A100, etc.) depends on how many compute units your account has. If your compute units run out or expire, the notebook will suddenly disconnect and turn into free tier with limited features until you renew compute units again.

[viral mason]

ew money

I hit those things who ask for money with a fly swatter

[hallow thistle]

Additionally, you can toggle "View resources" in "Runtime" menu to see how long your current notebook runtime lasts.

What's up? Do you have anything else about RVC or W-Okada or other AI program?

[flint vapor]

Hi! do you need support?

[young halo]

130 epochs in 2 hours of runtime, do you think i can get to 200 before it breaks?

[cursive juniper]

what to do? when i try to start using the ngrok server i get this error(google collab).

[viral mason]

https://tenor.com/view/get-yo-ahh-on-boy-gif-26030997

[brittle wing]

How do i get the vb cable louder without increasing the gain

[modest willow]

hello gamers, i humbly ask to see if there's been an update for the aicovermaker for the rtx 5000 series 🙏

[viral mason]

what is that?

[modest willow]

the aicovermakers an app for making ai voice model covers for songs, the rtx 5000 series are the latest gpus from nvidia

[viral mason]

ah oki, I wouldn't know about that but feel free to ask a mod or helper

[blazing mural]

is there any voice models that acly sound real..

like i cant find any that dont sound like mf siri

[simple ore]

colab?

[tidal root]

on rvc by w-okada it works good but i just have a issue where its glitchy/choppy and its hard to tell what im saying, how do i fix that

[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]

@viscid moss is there any update?

[tidal root]

6900xt, windows 10, make my voice clearer in rvc to where its not choppy/laggy, tutorial link: https://www.youtube.com/watch?v=SxdnGxicJOg&list=LL

it doesnt let me put a screenshot

[low shard]

that tutorial is outdated asf, it uses an over year old version of original wokada, along with vb audio cable that causes issues on windows

forget everything from video tuts, theres no updated video

-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 the wokada deiteris fork guide

[pseudo swift]

does anyone know where the tensorboard data is stored on google drive?

because my tensorboard bugged and stopped updating

[tidal root]

ty bro

[pseudo swift]

for this data

[tidal root]

so it wont be laggy for amd gpus

[pseudo swift]

can anyone help?

[simple ore]

use charts on the SCALARS tab

[pseudo swift]

is there a way to download it so i can get more precise numbers?

[wanton axle]

how to make my own voice ai model? pls😭

[simple ore]

scalars tab has big charts

avg_50 is the best option if you're using up to date software

[pseudo swift]

ohh okay ty im using applio's colab

[hazy eagle]

does anyone know how to avoid this

[modest willow]

i don't like the idea of using colabs because a lot of the models i use are private/made from people i know

[low shard]

it has better performance

[low shard]

what's ur pc gpu? operative system? and what do u want to do exactly?

[low shard]

what's ur pc gpu? operative system? and what do u want to do exactly?

what tutorial link did u use?

[hazy eagle]

windows 11 GTX 1660ti
im having that error when im playing. it likely happens when i have a lot of players on my screen

[low shard]

can you elaborate what game, what do you do exactly, what tutorial link

we are a general ai server

[hazy eagle]

gameloop then i play codm

[low shard]

there's thousands of AI programs

[hazy eagle]

i use forkVC

[low shard]

never heard of forkVC before, can you send a tutorial link? might be you're just using the wrong one

[hazy eagle]

fork voice changer

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

[low shard]

that's called wokada deiteris fork, please don't simply call it voice changer because there's thousands of programs that can be called that way

[low shard]

set graphics to the lowest, close everything other program in the background, set 1080p with 60fps capped

also try playing around with other emulators like Bluestacks or LDPlayer9

never heard of gameloop before

[hazy eagle]

its best emulator for codm

[low shard]

i mean if you say so, i dont really play codm

anyways, show also a screenshot of your wokada deiteris fork settings

[low shard]

i mean also the normal ones

[hazy eagle]

i know you'll say use rmvpe but, i just crepe_full_onnx because it gives more realistic

crepe_full is gpu heavy. but with onnx it kinda good

[low shard]

f0: rmvpe without onnx
you can optionally use https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay for less delay

extra: 2.7
chunk: just be sure its higher than the perf value while in game

[low shard]

crepe is very sensible to noise, and onnx is meant for non nvidia gpus 😭

dont listen to yt tuts telling you to use crepe, if you have

Hello, welcome here, this is our AI help channel, are you looking for help?

If so, please elaborate

? what zip file

this is a general ai server

we do thousands of programs

Please Elaborate:

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

that one is original wokada, which isn't suggested anymore since like a year almost

also, you need to elaborate the other info too

Are you trying to catfish like duckus as an e girl?

[flint vapor]

Another catfish guy went away hahah

Turn off your passthru, is working without passthru?

Can I see the interface of your voice changer?

Oh true, send me in dm and I will reply you here again

On this channel

Direct messages, you can send me as message on my discord profile 😁

The version you have is outdated of okada, now I will send you a guide for download the new one, if you will not understand some steps you can text here 😁

-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/

Oh wait first of all

Which gpu do you have ?

Oh, umm I don’t know if the performance of okada are so good with intel hd graphics 620, you need to put very high chunk also, in the guide you have to download the version for amd, cpu

I will send you download link

But I not guarantee that will work because gpu is integrated and a bit old

https://github.com/deiteris/voice-changer/releases/download/b2332/voice-changer-windows-amd64-dml.zip

If it will not work fine you can use the version on cloud and is also on guide, I can’t help you so much on cloud because i know only about local

Because it uses your gpu for voice conversion

In real time

After you download, extract the zip, open the folder, run the exe inside the folder

I don’t know which chunks and extra you have to put for the intel hd graphics 620 and if is compatible but I’m really concerned

[low shard]

You sure you don't got any other GPU? That's too weak to even get recognized, it's integrated graphics

[low shard]

Integrated graphics don't even get recognized

[flint vapor]

Yes true I wasn’t sure, i saw is about from 2016

And is even integrated

[low shard]

So, you got no other GPU?

Uninstall everything

It's just wasted time

Your PC is too weak

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)

[flint vapor]

Yes it will go very slow

Use it on cloud @coral gyro

[low shard]

You can try but it will be just useless, trust me

That gpu is almost a decade old

It wouldn't get recognized, and it would run only on your CPU

What's even your CPU?

You're cooked

The program will have extremely high ping that it wouldn't even be called "realtime" anymore

Seriously, it's not worth it to even try

Either buy a better PC or use cloud (remote good PC with limited free time)

Show a screenshot, you prob didn't click start

[flint vapor]

Yes @coral gyro sorry if I make you download but I’m not into voice changer, I wasn’t sure about gpu, but if you want try for today or limited time I recommend you on cloud also if better option is always run it in local with better gpu

[low shard]

Ofc the inference would appear, but it would lag too much to start and actually use it

!give-media-perms 1h @coral gyro

I told you lol

[low shard]

Also DW I know you're just learning

Dubbing AI is paywalled, it runs on cloud remote good PC, not on your hardware

[flint vapor]

Try it on cloud for now until you will not take better gpu

Oh there is in the guide 😁

[low shard]

#✨│ai-help message

[flint vapor]

Oh yes unfortunately I not using okada from a lot and I was always used the old version, I still a bit have to understand which gpu are compatible and good settings haha

I don’t think there is some better vc in realtime they all use gpu

[low shard]

there's none for free and open source better than this lol

[flint vapor]

There are normal voice changers like voice mod but is not AI

[low shard]

they all either use ur gpu (which is too weak) or are paid and on cloud

@coral gyro things like voice.ai, stay turn on 24h on your pc to use it to power their service

so yeah

either use wokada deiteris fork cloud kaggle

or get a better pc

there's really no other option

we are a general ai server, we do many programs, I can assure you that

you mean like voice mod? that's just 3 random free voice effects lol

without ai its just special effects, you won't be talking like "homer simpson" or whatever are you trying to do

btw, what are you trying to do if i may ask?

[flint vapor]

Depends from what you want to do

Like

Voice mod is mostly just for voice effects

Never give up, you can try it on kaggle or colab

[crisp crest]

2025-07-11 22:52:39.5099092 [E:onnxruntime:Default, cudnn_fe_call.cc:33 onnxruntime::CudaErrString] No valid engine configs for ConvFwd_
2025-07-11 22:52:39.5128597 [E:onnxruntime:, sequential_executor.cc:572 onnxruntime::ExecuteKernel] Non-zero status code returned while running Conv node. Name:'Conv_18' Status Message: Failed to initialize CUDNN FrontendD:\a_work\1\s\onnxruntime\core\providers\cuda\cudnn_fe_call.cc:99 onnxruntime::CudaCall D:\a_work\1\s\onnxruntime\core\providers\cuda\cudnn_fe_call.cc:91 onnxruntime::CudaCall CUDNN_FE failure 8: HEURISTIC_QUERY_FAILED ; GPU=0 ; hostname=DESKTOP-72N6F13 ; file=D:\a_work\1\s\onnxruntime\core\providers\cuda\nn\conv.cc ; line=225 ; expr=s_.cudnn_fe_graph->create_execution_plans({heur_mode});

with the cudnn frontend json:
{"context":{"compute_data_type":"FLOAT","intermediate_data_type":"FLOAT","io_data_type":"FLOAT","name":"","sm_count":-1},"cudnn_backend_version":"9.7.1","cudnn_frontend_version":10700,"json_version":"1.0","nodes":[{"compute_data_type":"FLOAT","dilation":[1,1],"inputs":{"W":"w","X":"x"},"math_mode":"CROSS_CORRELATION","name":"","outputs":{"Y":"::Y"},"post_padding":[32,0],"pre_padding":[31,0],"stride":[1,1],"tag":"CONV_FPROP"}],"tensors":{"::Y":{"data_type":"FLOAT","dim":[108,128,128,1],"is_pass_by_value":false,"is_virtual":false,"name":"::Y","pass_by_value":null,"reordering_type":"NONE","stride":[16384,128,1,1],"uid":3,"uid_assigned":true},"w":{"data_type":"FLOAT","dim":[128,1024,64,1],"is_pass_by_value":false,"is_virtual":false,"name":"w","pass_by_value":null,"reordering_type":"NONE","stride":[65536,64,1,1],"uid":2,"uid_assigned":true},"x":{"data_type":"FLOAT","dim":[108,1024,128,1],"is_pass_by_value":false,"is_virtual":false,"name":"x","pass_by_value":null,"reordering_type":"NONE","stride":[131072,128,1,1],"uid":1,"uid_assigned":true}}}
EP Error: [ONNXRuntimeError] : 11 : EP_FAIL : Non-zero status code returned while running Conv node. Name:'Conv_18' Status Message: Failed to initialize CUDNN FrontendD:\a_work\1\s\onnxruntime\core\providers\cuda\cudnn_fe_call.cc:99 onnxruntime::CudaCall D:\a_work\1\s\onnxruntime\core\providers\cuda\cudnn_fe_call.cc:91 onnxruntime::CudaCall CUDNN_FE failure 8: HEURISTIC_QUERY_FAILED ; GPU=0 ; hostname=DESKTOP-72N6F13 ; file=D:\a_work\1\s\onnxruntime\core\providers\cuda\nn\conv.cc ; line=225 ; expr=s_.cudnn_fe_graph->create_execution_plans({heur_mode});

with the cudnn frontend json:
{"context":{"compute_data_type":"FLOAT","intermediate_data_type":"FLOAT","io_data_type":"FLOAT","name":"","sm_count":-1},"cudnn_backend_version":"9.7.1","cudnn_frontend_version":10700,"json_version":"1.0","nodes":[{"compute_data_type":"FLOAT","dilation":[1,1],"inputs":{"W":"w","X":"x"},"math_mode":"CROSS_CORRELATION","name":"","outputs":{"Y":"::Y"},"post_padding":[32,0],"pre_padding":[31,0],"stride":[1,1],"tag":"CONV_FPROP"}],"tensors":{"::Y":{"data_type":"FLOAT","dim":[108,128,128,1],"is_pass_by_value":false,"is_virtual":false,"name":"::Y","pass_by_value":null,"reordering_type":"NONE","stride":[16384,128,1,1],"uid":3,"uid_assigned":true},"w":{"data_type":"FLOAT","dim":[128,1024,64,1],"is_pass_by_value":false,"is_virtual":false,"name":"w","pass_by_value":null,"reordering_type":"NONE","stride":[65536,64,1,1],"uid":2,"uid_assigned":true},"x":{"data_type":"FLOAT","dim":[108,1024,128,1],"is_pass_by_value":false,"is_virtual":false,"name":"x","pass_by_value":null,"reordering_type":"NONE","stride":[131072,128,1,1],"uid":1,"uid_assigned":true}}} using ['CUDAExecutionProvider', 'CPUExecutionProvider']
Falling back to ['CPUExecutionProvider'] and retrying.
^

what can i do

[flint vapor]

It will work ok

[low shard]

wdym with troll? sound like an e girl to catfish? or just sound like peter griffin? or just random voice effects like robotic?

[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

[crisp crest]

i cant send a screenshop

[low shard]

elaborate everything else i asked

!give-media-perms 1h @crisp crest

[flint vapor]

Very old version

-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]

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

[low shard]

all yt tuts are outdated for voice changers

[flint vapor]

Download the nvidia version

[low shard]

yep, follow what Kakao said^

[flint vapor]

And uninstall your current okada

[crisp crest]

is that the newst version

[flint vapor]

Yes 😁

[crisp crest]

wtf is that right

[flint vapor]

Is the interface

Means is all correct for now

Just first of all put your normal gpu

Instead of cpu

[worn silo]

I want to perform model quantization on qwen2.5vl:3b model, how can i do it?

Using huggingface or pytorch or smth realted

Can anyone please help

.

[mild sparrow]

how do i decrease the delay?

[flint vapor]

I think there are already quantized versions on huggingface, use GGUF

[flint vapor]

Not transformers

[worn silo]

why?

[flint vapor]

Is slowed and outdated

GGUF models are casters

*fasters

[worn silo]

u mean slow to give me the quantize model?

[flint vapor]

Oh no sorry I didn’t mean that but is slow in general like in general use

Even if is quantized

[worn silo]

oh shit

[flint vapor]

If I’m not wrong there are already quantized models on hugginface of this model, if you put in your profile your gpu and cpu

[worn silo]

the vlm will be slow even if i give it response?

[flint vapor]

Huggingface will estimate which version you should install

[flint vapor]

Yes, but, if you use ollama or Lmstudio will always download you gguf

[worn silo]

i am using ollama

[flint vapor]

Wait just lmstudio, ollama use their own format for models

[flint vapor]

But also GGUF

[worn silo]

also like i want to increase the output time of my model, so thats the reason i am doing quantization. cuz its the last thing i can do to make it faster

[flint vapor]

When you go on huggingface option of your profile, I’m not remember correct section but you should find some AI services as Ollama and you have to check the box

[worn silo]

ohh

[flint vapor]

I think you can find also quantized versions on huggingface

There are many GGUF

[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

[worn silo]

also so actually the ollama uses https api calls in the back to its local server to give the output. so is there a faster way?

[low shard]

yep
gpu: rtx 3060 ti
extra: 2.7
chunk: 80ms
f0: rmvpe without onnx
input: microphone
output: line 1
monitor: headphones optionally to hear urself

lmk

[flint vapor]

Oh there aren’t other methods for do faster way but only download quantized version as you said, for this model if you search on google the name model plus GGUF you will have a list of quantized versions

https://huggingface.co/Mungert/Qwen2.5-VL-3B-Instruct-GGUF

[worn silo]

got it

[worn silo]

ohh great i actuallt made my self

[flint vapor]

All quantized versions

[worn silo]

my god wow

thanks man

really apperitate it

[flint vapor]

Oh so nice! I’m not so much expert in how quantize model but I know there are many quantized models on hugginface, even the less popular models haha

[low shard]

GGUF models are goated

[flint vapor]

If you click use this model you will have ollama option and it will install automatic

[worn silo]

thats great man

[flint vapor]

Yes true 😄 transformers is a bit outdated if I not wrong , GGUF are very good and way faster

[flint vapor]

You welcome! If you need more help you can ask

[worn silo]

ur right

i got it

[worn silo]

you are amazing

i was taking me so long to do, u gave me the prefect result

[flint vapor]

Thank you so much ☺️

[worn silo]

Also mate

btw i am actually using this vlm for ocr extraction

I want to make the output more faster

do you have anything in mind that can do it?

[flint vapor]

Oh usually the output depends always from the quantized model you selected, but ollama should recognize image if you drop them, of course also the model have to support images

😄

[worn silo]

no that i get

but like from the vlm i am getting the ouput in 0.4s of the ocr

per image

anyway to make it lesser

[flint vapor]

Oh i think is depend all from the gpu, which gpu do you have ?

Unfortunately I’m not so much into vlm 😅 I hope you can forgive me, but usually all the times you receiving are from
Your gpu and the model

[low shard]

i mean they aren't outdated iirc, but GGUF are a suggested file format since you can choose depending on how good is your hardware

[worn silo]

maxed out, but need more faster

[worn silo]

np man

apperitate it

[flint vapor]

Oh yes true is not outdated but is always recommended to use GGUF

[low shard]

for example, I use flux.1-d gguf q8_0, on my rtx 4060 ti 16gb for comfyui, the results are 99% identical and its sligthly faster and uses less vram

[flint vapor]

You welcome 😄

[worn silo]

also like GUFF cant be cusiomized asper your requiremnets

so will need to qunatize your self

using pytorch or transformers smth

so exactly asper your requirement

[worn silo]

Do you know about vlms, ocr and stuff?

[flint vapor]

Yes true I used to have ultrarealfinetune the full one, then I switched to the quantized version and is way faster and same results at 99%

[low shard]

about ocr, i know about EasyOCR and Tesserract but haven't used them in a long time

[flint vapor]

Maybe you can resize a bit the image or use lighter prompts but I’m not sure this will make a bit faster the output haha I’m only saying some things but maybe the things I’m saying are a bit nonsense

[worn silo]

no it makes sense, i have already did that optimization and it becasue faster when i used smaller prompt and also on cropped portion of the image

[flint vapor]

Nice! I think it mostly depend from the quantized model in particular and from gpu usage if I’m not wrong 😄

If you do ollama run (your model) —verbose

It should give you full details of what is happening during the generation

Like timings and cpu and gpu usage

[worn silo]

bro

thanks

will def use that

[flint vapor]

You welcome ! 😁

[worn silo]

i didnt know about verbose

[flint vapor]

Oh yes is a command a bit unknown is used only for know details and what is happening while using the model 😁

[worn silo]

ahh ty

[flint vapor]

You welcome !😄😃

[jaunty bloom]

Is applio not really working for anyone else or is it just me?

[lusty ore]

can some1 help me

my voice sounds robotic but i have every setting really high and an really good gpu

[low shard]

Elaborate:

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