✨│ai-help

[low shard]

delete everything you got off it and read the guide i sent you, video tutorials are outdated

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

[timber bramble]

Do I use Mainline or the other one

[low shard]

use the 1st link, wokada deiteris fork

be sure to read it all

if you skip parts, you're going to have issues ofcourse

[low shard]

@timber bramble let me know

[hallow thistle]

There's no such a thing named "Mainline" in W-Okada. The actual mainline in W-Okada is bascially the original version of it, which is not really recommended to download and use so.

[low shard]

@timber bramble I meant this btw

[hallow thistle]

If you got confused by "RVC > Local > Mainline" at the left side bar, it's a regular RVC program where you make AI cover, not RVC "Voice Changer".

[shrewd stag]

It says Intel(R) Iris(R) Xe Graphics

Sorry for late response internet was out

[shrewd stag]

I'm on laptop

[knotty moth]

only cpu mode is viable

[shrewd stag]

I don't know what these words me

😭🙏

[low shard]

cooked, weak integrated graphics

are u sure u dont got other gpus?

[shrewd stag]

Damn

[knotty moth]

if you dont afford the processing time, there are cloud solutions like huggingface or colab to search

[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, but 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

so, either:

• use cloud (MOST SUGGESTED) which is using a remote good pc for limited free time
• buy a better pc
• wait a shit ton of time to run it on cpu

[shrewd stag]

😭🙏

Crazy

[low shard]

just use cloud, it's free

u can't expect ai to run on poor hardware

chatgpt doesn't run on your pc, it runs on cloud

[shrewd stag]

Google?

Where do I get dat

Sorry gang not a big pc guy

[low shard]

I gave you every single link above #✨│ai-help message

everything works except mainline

[hallow thistle]

That's an integrated GPU, which is never supposed to be used in AIs.

[charred briar]

does anyone know how to train a separation model? i want to make one so i can isolate this specific artists music cleanly by training a model off their official instrumentals and vocals

[hallow thistle]

The custom stem separation model? I see UVR5. But I've not seen anyone here ever done that.

There are plently of UVR5 models available for use extract audio.

[charred briar]

i just have a very specific type of music that a lot of models arent exactly made for

so i get very poor isolations

[viscid moss]

https://docs.google.com/document/d/17fjNvJzj8ZGSer7c7OFe_CNfUKbAxEh_OBv94ZdRG5c/edit?usp=sharing

It's hard

[charred briar]

i figured it would be

[viscid moss]

On that guide u can found how to make a separation model

[knotty moth]

training separation models ideally needs a huge amount of dataset

[viscid moss]

I recommend u to train a Mel Band Roformer one

[viscid moss]

[charred briar]

ye melband roformer is amazing

ive gotten some of the best results with it

[charred briar]

i have about 3 hours

of data

[knotty moth]

we've got good enough voc-inst models, so no more

[charred briar]

ive been using the one on mvsep,

[knotty moth]

however you could consider trying to get some guitar dataset and train on it

[brittle wing]

how to import a voice model to the server?

[hallow thistle]

You trained your own voice model?

[brittle wing]

im using a premade

[hallow thistle]

Do you wish to upload it to #1175430844685484042 or something else?

[brittle wing]

no i mean like upload it to the voice model server

[hallow thistle]

Sorry, but if it's not #1175430844685484042 in this server, then I'm not sure what you mean by "voice model server" then.

[brittle wing]

i mean like uh

uploading the voice model to the ngrok server thingy to run on cloud

[hallow thistle]

Do you mean like W-Okada? At least just remember and say the name of the notebook you're using.

!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]

ok so basically im using ngrok right and i cant figure out how to upload a voice model to it

[hallow thistle]

How many thing you don't know? ngrok is a website that gives token from your account to be used in a cloud service project, so that the program can display its UI on separated ngrok website.

Well, guess I have to tell you how to "upload a voice model into W-Okada" then. Telling every step by step is crazy.

[brittle wing]

i js need a quick tutorial on where to upload the index and pth file

[hallow thistle]

On W-Okada. There's a red arrow pointing to edit button, this is where you upload RVC voice models.

[brittle wing]

alright

[hallow thistle]

My guy, this is the quickest tutorial I can do.

You click on edit button, there's this popup. At the red arrow, there's an upload button.

Oh, before you upload voice model, make sure you have extracted one or two files (.pth, .index) from a voice model zip into a folder. Upload these files, and click "upload" button.

[brittle wing]

got it

[hallow thistle]

The instruction for W-Okada on both locally and cloud service is pretty much the same. The difference is where you run the program from.

[twilit forge]

whats the audio ai upscaler link the google colab version

[true quail]

I got a question for u

[low shard]

Wdym

What's your PC GPU? What do you want to do

[low shard]

Btw lmk since I told ya

[dark ginkgo]

I'm getting a widely different and unusable responses from anything llm compared to just using lm studio

the temperature is the same

[mossy gull]

@low shard Sorry for the ping , Just let you know that I got my prefeed result with "TTS" library!
As for the robotic voice i don't have to integrate RVC at all, I just used some voice filter with "signal" library to make it robotic and the client really liked it.
Thanks for the help yesterday, I really appreciate it 🙂

[low shard]

Yw

[waxen pike]

tell me the settings that will make the voice as real as possible. And tell me which model is good for Russian or Ukrainian language?

[true quail]

Russian I would said

[waxen pike]

go

[waxen pike]

pls help

[true quail]

Idk what u doing sorry

I thought you were asking if it's better Russian or ukran language

[waxen pike]

@low shard help tell me the settings that will make the voice as real as possible. And tell me which model is good for Russian or Ukrainian language?

[low shard]

uncheck sup1

uninstall vb audio cable

get vac lite from the same program guide

then get a good model and play with the pitch

[waxen pike]

Can you please tell me where I can get a quality model? You know the kind, maybe streamers are famous.

[low shard]

the only way is to check the sample and test them yourself

[waxen pike]

is this it?

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#models-to-try can help

[low shard]

yes, but uninstal vb audio cable

[waxen pike]

???

[low shard]

show a screenshot of ur wokada

[low shard]

read it

[waxen pike]

I'm not quite sure why I'm doing this and that it's probably a translation difficulty. I will now find information on how to remove the cable and all that is left is to find a quality model

[low shard]

of ur wokada

the literal program

for realtime voice changing in calls

[waxen pike]

I downloaded several models of the original of which Russian language but still the quality leaves much to be desired. Besides pitch should I touch something else?

[twilit forge]

is it starting to over train yet

[orchid flint]

No

[low shard]

you can try enabling force fp32 mode on in advanced settings, but it really depends much on how good was the trained model + the way you play with the pitch

[restive tendon]

any ai tools to transcribe a video for me? I need the full script of what I said in it

[silent stratus]

@glacial pollen is there any guide on finetuning melroformer models

[glacial pollen]

You'd have to ask on audio separation discord really

they are the masterminds and lots of people there who actually trained most known models

[silent stratus]

gochu thanks

u got a link

[glacial pollen]

one sec

hvgmU2yA4u

here

[silent stratus]

thanks a lot

[glacial pollen]

No prob man, hope you can find some answers

✨

[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.

[torpid oriole]

Hey I have a question in my mind that i use live voice conversion on windows using nvdia gpu card and its working perfectly , my question is - windows take more process than linux so if i use linux than it can possible i will get less delay on that cause linux take less compute to run that os in same nvdia card and spec , i need conclusion in same specification the rvc will work more faster than the windows one? (using same specs and same nvdia gpu card)

please response....

@low shard

[austere siren]

AMD rx 6600 and i want to use the voice changer in real time just for communication

[austere siren]

"! C:\Users\Blick\Downloads\voice-changer-windows-amd64-cuda.zip.001: Unexpected end of archive"

when i try to open the wokada deiteris fork file

[glacial pollen]

corrupted archive

also try to unlock the archive
( right click, unlock zip or however it's named there for you )

[outer moon]

Can someone walk me through how to download this i'm trying to download this and catch predators

[glacial pollen]

💀

[outer moon]

I acutally do that shit can you help me or nah?

[glacial pollen]

Uah, nobody asked for a hero ™️

We provide support for AI, we don't ask you for ideals or reasons you need it working

so spare that for yourself if you may

[outer moon]

Cool

[glacial pollen]

-rt

[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

[glacial pollen]

cool

now that we've sorted it out, there's the link

[glacial pollen]

Ps, it doesn't hurt to say " thank you " my dude ✨

In case of troubles, @ nick008 and others can help

[outer moon]

Nah after that hero shit nah

[glacial pollen]

It's facts.

[low shard]

-realtime

1st link

[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

[glacial pollen]

I hate the direction rvc is taking in most people's eyes

[austere siren]

i downloaded everything it just sounds so robotic and idk which settings are good for my pc

[outer moon]

Well idk what other intentions its made for lol

[low shard]

Have patience

[low shard]

Show a screenshot of ur wokada

!give-media-perms 1h @austere siren

[austere siren]

got you

[glacial pollen]

fun and entertainment, not catfishing or predators hunting.
Not in my eyes, at least

take it or hate it, I won't change my attitude

[outer moon]

Well we are doing something better with it rather then dwelling and chatting with strangers

[glacial pollen]

Shit should be left to proper institutions and police

[low shard]

Well technically Linux uses less resources and could result in less delay, tho I don't think there's any helper using Linux there

[glacial pollen]

Fair enough, but for future, legit spare the details

[outer moon]

Ah I see why

[glacial pollen]

it's not like it's gonna increase your chances of getting help

[outer moon]

Nah gfy

[glacial pollen]

Alright kiddie

go enjoy your hero-play

[modern swallow]

why every time i use virtual cap to use rvc on discrod my laptop loses sound

[glacial pollen]

virtual cable?

[modern swallow]

Yeh

[glacial pollen]

well, did you route it properly?

voice changer -> vb -> whatever uses it
main line of sound should stay unchanged

[low shard]

Chunk: 192
Extra: 2.7
If you're planning to use Server to have less delay, use wasapi https://rentry.co/LessDelayWasapi

Also be sure you got a good model and play with the pitch

[glacial pollen]

so ig unless you screwed up the routing or chain, idk

[low shard]

Vb? Vb audio cable gives issues on windows

It's reported by users that it randomly stops working, so we always suggest vac lite instead

[glacial pollen]

then vac

ye I use em both actually ( work just fine

my b

but perhaps cause of my drivers being a lil tweaked

[low shard]

Elaborate more:

• your PC GPU
• what tutorial link did u use
• what's exactly the issue
• a screenshot of ur wokada

[modern swallow]

Im not sure either let me check again :v

[low shard]

!give-media-perms 1h @modern swallow

[glacial pollen]

In any case, make sure you use what Nick wrote, the vac lite

then you wanna route the voice changer's output to:

[low shard]

And be sure to be using Wokada deiteris fork

[glacial pollen]

and if you wanna hear yourself

[low shard]

Not some random YouTube tutorial

[glacial pollen]

you wanna get this lil helpful thingy

[low shard]

You don't need that to hear yourself

[glacial pollen]

it's for lowest latency

[low shard]

There's an option called monitor in wokada deiteris fork

You can set it to your headphones

[glacial pollen]

is it also ultra-low latency?

[glacial pollen]

cause it gives nearly live-monitoring delay ( idk, in mics / interfaces

[austere siren]

will do thank you

[low shard]

It's pretty good, have you tried Wokada deiteris fork?

[glacial pollen]

ye

but I sticked to mme

as you can uhm

you can get the delay so damn low

[low shard]

I personally never had any delay issues with wokada deiteris fork monitor option

[glacial pollen]

by proper tweaks

[glacial pollen]

tho ye u right

if simpler solution exists, better recommend that

[magic belfry]

I can't run server file its a .spec

[low shard]

Lemme ask in staff chat if other staffers ever felt like monitor had delay issues though because I personally never felt that

[glacial pollen]

it ain't about issues tho

but rather actual delay

latency

I'm just very sensitive to it, that's really all there is to it
( am an ex osu player so audio latency is important as hell for me

[low shard]

Yeah it's better to ask if others tried that program too and felt it had less delay

[glacial pollen]

Probs not

not a lot of people know of it

quite obscure on the web, in ' monitoring ' purposes like that

[austere siren]

this helped so much thank you

[glacial pollen]

anyway, you can guys test it out and see if you like it
( careful, you can potentially trade cpu for latency

[magic belfry]

Anyone able to help me

[low shard]

Elaborate

[magic belfry]

I downloaded the file extracted I see no exe to run

[low shard]

• what's your PC GPU
• what do you want to do
• what tutorial link are you using
• what's the issue

[low shard]

There are thousands of ai programs and files you could be talking about right now

You need to elaborate

[magic belfry]

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

[low shard]

@magic belfry please reply to all this

I know that's the right tutorial link for realtime voice changing, Wokada deiteris fork, but I need to know all info to help you out

[low shard]

Never played OSU honestly

Nor geometry dash

[magic belfry]

3080
Have a working girl voicechanger
https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/
Can't find the exe to run or setup the thing

[low shard]

Alright, did you download the windows Nvidia version and used WinRAR/7-Zip/Peazip to extract it?

[magic belfry]

Yes

[low shard]

When it's extracting, there will be 2 .bat files and then a MMVCSERVERSIO folder, you need to open the only folder there and then there will be a MMVCSERVERSIO.exe

[magic belfry]

I see 4

[low shard]

Show a screenshot

!give-media-perms 1h @magic belfry

[low shard]

Did you git clone the repository

I feel like you went on the GitHub repository and either git cloned it or clicked the download as zip

Delete what you got, read the guide link you sent, there will be the already compiled version there

Remember to not skip steps

[glacial pollen]

btw nick, if you can give this dude a warning, I'd be grateful ^

Ain't happy seeing some kids going around gfy'ing around

[magic belfry]

I just joined after seems you were a tad rude when he is just trying to help people and you made a side comment.

[glacial pollen]

Nah, I was actually quite reasonable

• ^ to extend my view on this clown-fest

I don't want software I care about and I work on to be associated with such practices
rvc / applio should be associated with fun and creativity, not self-claimed justice heroes

And no shitty " mr go f yourself " gonna make me think any different.
If you ask for help, you ask for help, we ain't asking you why or for what purpose.
If one needs to show their ego on how important is their goal, then that's seriously fked up.
( cause that's the kind of vibes I got from the dude )
And most importantly, regardless of what's going on, there's no excuse for going as far as the dude allowed himself to.

[magic belfry]

saying its corrupt

[outer moon]

I think bro is sensitive

[magic belfry]

agree a tad

[glacial pollen]

And I think bro's full of himself sending his minions on us
sad but you do you kiddie and be nice

[modern swallow]

it works pretty well on discord with me calling to another device but my laptop itself has no sound :v

[magic belfry]

? I need help i'm not with no one

[glacial pollen]

whatev, redownload the archive and try again

[magic belfry]

Rude

[glacial pollen]

xd

I said redownload the archive and try again

[outer moon]

Wsp with you saying Kidde so much and saying all this hero stuff and how it ain't cool is there something going on?

[glacial pollen]

And for future

!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.

[glacial pollen]

Cause you def act like a kiddie

talking 'bout being sensitive ay

[outer moon]

You like talking to kids or sum?

[glacial pollen]

You're legit out of mind

[outer moon]

Not everyone is a kid pal

[glacial pollen]

@crude flame

come over see this shit

I can't

[outer moon]

👐

[glacial pollen]

You surely act like one by going all sensitive on my opinion on what you do

[crude flame]

Lol

[glacial pollen]

and yet spitting out " sensitive " into my face, fkin hypocrite-act there

fr

I'm 24 yo and too old for that kind of shit. Get out of my face man. do your stuff, topic's done

[magic belfry]

Its saying its corrupt still

[glacial pollen]

Perhaps try to use 7-zip instead

also, where did you download the archive

[magic belfry]

https://github.com/deiteris/voice-changer/releases

[glacial pollen]

a

[crude flame]

You gave me a good chuckle :thumb:

[crude flame]

AI hub is a server

[glacial pollen]

😌 🍷

[glacial pollen]

oh, now I see what you did

because these are parts

you need both, then unpack from 001 ( as in, open up that archive and unpack

whenever you see an archive with " 001 " and such

means it's parted and you need all to unpack it properly

[magic belfry]

ah

[glacial pollen]

Yeah, happens, dw

[magic belfry]

Aw thanks

[glacial pollen]

Np, hope it goes well for ya

[modern swallow]

Can I use a voice meeter so that the sound can still go through the virtual cable and I can still hear it?

[strange mountain]

does it work in vietnamese voice?

[austere siren]

same problem

[low shard]

never skip steps

@austere siren @modern swallow uninstall vb audio cable and read https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable

[outer wasp]

Hi! How to make voice not so robot and say the exact word?

[low shard]

Elaborate

• what's ur PC GPU
• what tut link are u talking about
• what is the exact issue

There's thousands of ai programs

[bold zinc]

hello gys,
what are the best techniques to improve the accuracy of a classification model (tabular data with alot of categorical variables)

[unreal linden]

Hello, i want to know which f0 det is good, rmvpe,crepe or fcpe?

[paper lotus]

Is there any way i can make my own ai voice model

by just using a clip of them talking

[low shard]

Yeah, what's your PC GPU?

[paper lotus]

rtx 4070

[paper lotus]

also what should my input and output device be

[low shard]

Rmvpe is the most robust to noise and good quality, crepe is least robust to noise but soft, fcpe is not heavy and soft but not as accurate as rmvpe

[low shard]

As you got a good PC, you can use RVC locally, you can choose between:

• Applio: A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
• Mainline: The original RVC

[knotty moth]

it is more than enough for training

[low shard]

In wokada, input should be microphone and output should be line 1,

In other programs, input should be line 1 and output ur headphones

[knotty moth]

mostly you'd need to train (finetune) on a single speaker

[paper lotus]

oh ight

[paper lotus]

whats finetune

[paper lotus]

applio would be better right?

[low shard]

Yes

[low shard]

Also, be sure to use wokada deiteris fork, and not video tutorials

For best quality and performance

[knotty moth]

training using a base pretrained model

[vague edge]

the voice changer is working well but as i finish my sentence it sounds robotic and strange

[low shard]

Elaborate:

• your PC GPU
• what tut link did u use
• a screenshot of ur wokada

!give-media-perms 1h @vague edge

[low shard]

Lemme guess, you used a video tutorial?

[vague edge]

yeah lmao

[low shard]

Because you're using original wokada which is worse in quality and performance

[vague edge]

i just found the link of the github on their

[low shard]

And the settings are completely wrong

[vague edge]

i thought so because on other peoples screenshots their ui is different

sorry

[low shard]

Video tutorials are extremely outdated for wokada and rvc

Uninstall all you got off it

-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

[vague edge]

yep

thank you

[low shard]

1st link, Wokada deiteris fork

Better in quality and performance

[low shard]

Dw

[low shard]

Yw

[paper lotus]

is chunk the quality or is extra the quality of the ai

im confused

[low shard]

Chunk controls the delay, extra controls the quality

[paper lotus]

what setting should i use

[low shard]

Share a screenshot of ur wokada

And tell ur PC GPU

[paper lotus]

i just did bro

[low shard]

I meant a screenshot of the program

To be sure you're using wokadaa deiteris fork

And not some old YouTube tutorial for original wokada

!give-media-perms 1h @paper lotus

Also so I can check all your settings

And be sure to give the best ones

[knotty moth]

try these settings https://rentry.co/forkvoicechangerguide#known-working-settings-for-chunk-and-extra

[low shard]

We need to check before that they are even using wokada deiteris fork 😭

Else the settings would be different

[vague edge]

Just quickly making sure is this the deiteris one? https://github.com/deiteris/voice-changer for nvidia?

[knotty moth]

well I mean the version being used like^

[paper lotus]

i was just testing out the one thats hosted in web

its acting weird

[low shard]

Yes, be sure to get the download link from the guide which is precompiled, because the one on GitHub is divided in 3 zip parts because of the file size limit

[vague edge]

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

oh sorry this one https://huggingface.co/Shadicti/deiteris-Fork/blob/main/voice-changer-windows-nvidia-b2332.zip

[low shard]

f0: rmvpe without onnx
Extra: 2.7 for not having cutoff issues
Chunk: 128
You can optionally use server for less delay which is explained in the guide
Also you can optionally use force fp32 mode on in advanced settings for more delay but better quality and stability

[low shard]

Yes

[vague edge]

@low shard it still sounds a bit robotic at the end, heres my settings.

[low shard]

Install vac lite from the same guide
Set extra to 2.7
Chunk: 192

[strange mountain]

web version or mcvv version which is better? When I speak, it's like sometimes it's silent, sometimes it's noisy, sometimes it makes some kind of sound. Even if I hear the sound of the keyboard, it still accepts it (I speak Vietnamese).

☠️

[low shard]

It isn't called web version/ mcvvitd commonly called wokada deiteris fork there

Share a screenshot of the program

[strange mountain]

ok

[low shard]

!give-media-perms 1h @strange mountain

[strange mountain]

you mean the program settings right?

[low shard]

The whole program yes

Be sure to get a better model and play with the pitch, enable sup2

You can optionally use force fp32 mode on in advanced settings for better quality and stability at cost of more delay

[strange mountain]

oh ok

[strange mountain]

can i ask ? thank you, also is it coming from my mic (to the quality of the sound)

[low shard]

A decent microphone should be fine for wokada

Unless you have a 10 cents microphone off Temu

[sinful shore]

I am trying to merge two checkpoints using Supermerger in Automatic1111 or ForgeUI, and I discovered that one of the checkpoints, which I want to merge, does not have any metadata so the merge errors out. Is there a way to merge checkpoints with no metadata?

[strange mountain]

it really works but the sound is very low is there any way to fix it?

[torpid oriole]

ah okie

than i will try that and if theres any good changes appear i will inform you guys thank you

[low shard]

Elaborate

• your PC GPU
• what u want to do
• a screenshot of ur wokada

[low shard]

Yw

[mortal peak]

So guys

Does the gtx 1650 work on the ai voice thingy

While playing

Like does it play it so good

While dc is open

[grim perch]

what is the best way to clean audio for a model

and whats the most effective model creation tool

[olive plover]

Hi, I feel that the sound is unrealistic and not beautiful in FORK

Arabic dialect

@low shard

[olive plover]

What should I do, please? When I enter into a discussion or talk with someone, I feel that the voice seems unreal and beautiful.

[olive plover]

Are there any settings I should change to make it look better?

[simple ore]

chunk is wrong for 4070 i think

[olive plover]

Why how much should I put it?

[olive plover]

@low shard

[simple ore]

128 ms chunk + 2.7s extra for 4070

read https://rentry.co/forkvoicechangerguide#finding-my-own-settings-for-chunk

[olive plover]

I did that and the sound got worse and faster

[simple ore]

for what/

[viral mason]

it does this every time

only on specific files

[viral mason]

like this

[knotty moth]

mostly light games like roblox

the fork version should be capable for that https://rentry.co/forkvoicechangerguide

[grim perch]

im on the ai hub doc

[viscid moss]

out of quota

check ur quota on ur profile

[grim perch]

ohh thank you for the upload instructions

[viscid moss]

move to Colab/Kaggle or local if still

[viral mason]

literally am not 😭

[viscid moss]

out of quota?

[viral mason]

I showed this in the video I think

[viscid moss]

lemme test

[viral mason]

wait no I didn't

[viscid moss]

and.... it's working for me

same model and output format as u btw

I think is kinda bugged

I've seen that problem before

[viral mason]

could u send the output

[viscid moss]

DM

[viral mason]

sure

[viral mason]

is there a fix?

[viscid moss]

nope, just wait and see

try logging out and then logging in again

try without logging too

[viral mason]

alr

[hollow beacon]

i got a question that i feel like is probably asked alot

i manage to learn how to setup the realtime rvc voice but like how do i get it to connect to other applications like a recorder, i want the voice changer to work while i voice chat and game

as in i want the voice changer to work realtime as im playing aswell, where the voice changer can be heard by other people

[knotty moth]

generally we recommend fork wokada over the rvc realtime

https://rentry.co/forkvoicechangerguide

[hollow beacon]

i am now more confused, where do i get the fork version?, from the github or from the guide website

[knotty moth]

you haven't read the guide quite enough, madoka

[low shard]

No need to ping again, have patience

[low shard]

Force fp32 mode on

Try a better model

Play with the pitch

Not all models are good

[low shard]

Btw don't share that link, we are using the docs link now

As of right now it should have the same things but not sure if it will change

[low shard]

What's your PC GPU?

[hollow beacon]

amd radeon

[low shard]

Which

[hollow beacon]

ryzen 7000

[low shard]

That's a CPU

[hollow beacon]

mistook that

[low shard]

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

The GPU is extremely important

Since it's needed for all heavy tasks like AI

[hollow beacon]

amd radeom (TM) graphics

[knotty moth]

well I was too away from checking the updated docs

[low shard]

That's cooked, weak integrated graphics

Do you have any other GPUs?

Like GPU 0, GPU 1

[low shard]

It's the first link in -realtime

[hollow beacon]

none at all unfortunately

[low shard]

Then your PC is extremely too weak to run it

Your only options are:

• buy a better PC
• try CPU mode which is unstable
• use cloud (remote good PC) with limited free time

@hollow beacon Kaggle is a google cloud computing service that offers 30 hours weekly of good GPU if you verify your phone number, that should be enough free time for you right?

[hollow beacon]

ill find other plans, for now thanks for letting me know

[low shard]

Wdym?

It's free

[grim perch]

does that innitial dip suggest overtraining?

[torpid oriole]

can you suggest a good linux os for using live voice conversion using nvdia gpu card? 🙂 @low shard

i tested on few os where nvdia drivers are giving error to download

[low shard]

I don't really use Linux, maybe Ubuntu or Mint

I wouldn't really change my whole OS just for trying to get lower delay on a single program

[torpid oriole]

i dont have any problem i use my spare pc for this and i just wana check so suggest me ubuntu or mint?

[low shard]

They are pretty easy to use iirc yeah

[knotty moth]

he's trying to get it working with Nvidia gpu driver under linux

[torpid oriole]

yess

no worries let me try in ubuntu

[strange mountain]

it works but it's small, my gpu is still working fine, i just want to fix that. 💀

[wanton saffron]

need help

6700 xt should be good enough

so im confused

[grim perch]

hey im having some serious issues with this same version of the rvc voice changer. why is it that 90% of the time when i launch it it just doesnt work.

when i press start. nothing happens.

im abt to lose my mind

[wanton saffron]

it only works when i set the chunk to max

before it worked fine

[grim perch]

like the ms line just doesnt even start

its like i never hit start

see its started but nothing is happening

[wanton saffron]

i think its the extra

its buggy for me

try setting chunks higher

[low shard]

which is why i’m asking to you to elaborate

to check ur settings

[low shard]

F0: rmvpe onmx

Output: line 1

what program are u using it with

[wanton saffron]

deiteris

version of okades

[low shard]

set extra to 2.7, uninstall vb audio cable

[strange mountain]

uhm so every time i speak the sound still works fine yea and the sound is very low i tested it several times and read the Docs (still cant fix)

[low shard]

yes im asking which program are you using along the program, like discord vc or a game

[wanton saffron]

gpu has more than enough to handle the task

[low shard]

Try to set in and out to an higher value, if it still doesn’t work, show a screenshot

[wanton saffron]

i'm doing this withougt games rn but its fucking up somehow

[low shard]

check the triangle

amd gpus have to always check it everytime they change settings

the triangle will save ya 🙏

[wanton saffron]

this?

[low shard]

yes

[wanton saffron]

i'll try to do it

[low shard]

you have to always do that when u change settings since ur on amd

[wanton saffron]

alright thanks

also do i need to use virtual cable for this?

[strange mountain]

same

[low shard]

A VAC (Virtual Audio Cable) makes a fake audio device, used to re-route the audio of different programs

In Wokada context, it's used to get the output of wokada as the input in other programs

TL;DR: yes

[wanton saffron]

alr thanks

[low shard]

share a screenshot of ur wokada pls

[strange mountain]

cant 💀 drop

[low shard]

!give-media

[wanton saffron]

still havin massive lag issues though

idk how

[low shard]

!give-media-perms 1h @strange mountain

[wanton saffron]

idk whats wrong anymo

[low shard]

set chunk to 200, f0 to rmvpe with onnx

click stop if it doesn’t let u edit it

[wanton saffron]

extra?

how much could my pc handle without it bugging out

[low shard]

it should be able to handle it

[wanton saffron]

max?

[low shard]

set also the in a bit louder

[strange mountain]

output right?

[low shard]

i mean the in too

[strange mountain]

oh ok

same

but still more big

also small

[grim perch]

hey where do i download the client?

[grim perch]

this version

i cant find the download

[wanton saffron]

https://github.com/deiteris/voice-changer

[grim perch]

god damn 😭 \

gonna be here till 2029

the error that was causing it not to work has to be here

[knotty moth]

for server mode, set the sample rate to 48000

[grim perch]

unfortunately thats what its at :/

[knotty moth]

also in control panel perhaps

[grim perch]

yeah its all on 48000 i have no clue what could be wrong

it was working fine yesterday

ok i saw that it was my headphones that were breaking it

i just took my headphones out of the output and it booted up

wtf

i cant hear it now but at least its working man

[low shard]

is ur audio output low only when using it in a call?

what program are u using it with?

because some calls like whatsapp reduce the media volume

[torpid oriole]

<@&1159293204038955078>

[strange mountain]

no all

[strange mountain]

no i didnt use anything reduce

i turn off all

[low shard]

they do that by default

maybe check https://www.alphr.com/windows-reducing-system-volume/

[strange mountain]

so i think i will turn off do nothing in line 1

[low shard]

else, try increasing in and out again

[low shard]

it’s a thing done by the other call programs sometimes

[tranquil radish]

Does the gtx 1650 work on the ai voice thingy While playing Like does it play it so good While dc is open and roblox

[strange mountain]

uhm what should i chose

[autumn pewter]

where i can find anyother voice for voicechanger?

[tranquil radish]

#1175430844685484042

[autumn pewter]

❤️

[formal ermine]

why is my res so high, like the voice changer starts being delayed even more when i try to play games

[primal nimbus]

Hello.
Could I please have a link to the current version of AI Hub?
Thank you

[sharp monolith]

hey just a quick question, i want to train a text-to-image model without using any pretrained models like SD, anyone know a good repo that helps out with that?

i've prepared a dataset for it

[primal nimbus]

Would you help @sharp monolith ?

[sharp monolith]

I'm not sure what you mean by "current version of AI hub"

[primal nimbus]

Need a program for Voice Modulation, currently I have some very old Voice Changer Client Demo :/.
And I would like the current newer version , look at the screenshots of other people above

[primal nimbus]

such as:
My version is: v.1.5.3.18a

[low shard]

that’s outdated asf

over a year old version

[low shard]

ai hub is not a program

[primal nimbus]

And would you help me to have the latest version ?

[low shard]

RVC = Retrieval-based-Voice-Conversion, the best 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)

you need reaaltime voice changer?

tell your pc gpu

all video tutorials are outdated

[primal nimbus]

It needs this program what the above people in the screenshots have, could you help?

on this screenshot

[low shard]

That’s a realtime voice changer for calls, specifically wokada deiteris fork

[primal nimbus]

13th Gen Intel(R) Core(TM) i5-13600KF 3.50 GHz

[low shard]

is that what u are looking for

[low shard]

that’s a cpu

[primal nimbus]

Sry, GTX 4070 TI

[low shard]

i think you mean rtx, yeah it would be good enough

delete everything you got off youtube

-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 the guide

[low shard]

you don’t see any settings about the device sound reduction? You sure you followed that guide link i sent

[low shard]

should

[low shard]

show a screenshot of ur wokada

!give-media-perms 1h @formal ermine

[primal nimbus]

And in this program as above, you can insert a voice from the voice -models category and it will work properly ???

[low shard]

it will work as a realtime voice changer for calls/games, not pre-recorded audios

Is that what u are looking for

[low shard]

read this to understand it better

[primal nimbus]

Okey that's what I'm looking for
That it would change the voice in real time based on just these voice models

[low shard]

then yes, read the guide i sent

[primal nimbus]

Okey

Currently I have played Python 3.10.11 and Okada but the program does not start :/.

[low shard]

you don’t need python

Delete that python version + the wokada you got off youtube

forget everything you got off youtube

Read the guide, it’s everything there

[knotty moth]

no pretrained model to use means you'll ideally need all the image data around the internet to get acceptable results

[primal nimbus]

Can you send me a direct link on priv message? Not lying I am green in this and just learning

[knotty moth]

that's the reason to do finetuning mostly

[primal nimbus]

In short, how do I download a new version of this? So that I click and it turns on and that's all it needs ^^.

[sharp monolith]

Oh I'm training it on something pretty vague, I don't really want other images as data for the model

[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]

you need to read the 1st link up there

[primal nimbus]

and what do I download from it ?

[low shard]

that’s the only way to get it

[knotty moth]

the pretrained model provides building blocks for everything you need

[low shard]

the vac lite and wokada deiteris fork for nvidia windows, read it all and don’t skip important steps

it isn’t a 1 click installer or as easy as chatgpt, but all you need is inside that guide

[primal nimbus]

Okey and now when I went into the guide I have the VC Client version

[low shard]

vcclient is the old original wokada

delete it

delete all you got off youthbe

[sharp monolith]

Hmm... Idk I'm training on some pretty slow hardware and I'm tryna make this as lightweight as possible

[low shard]

you need only the things in the guide

[knotty moth]

lora training is an approach to do besides the regular finetuning

[primal nimbus]

Okay but I'm not some kind of professional either and this version is totally enough for me just the key word on what to go into to download it.

this from a screenshot

[sharp monolith]

:(

[low shard]

please don’t keep replying that message since it keeps pinging the user, to have that version, you have to read the guide

there’s no other way to use the program if you don’t read the guide unless you want to do a mess

that other user read the guide too

it isn’t as easy as just downloading it and clicking start

[primal nimbus]

Well, I'm off to read the instructions

[primal nimbus]

😦

[primal nimbus]

This is the version I currently have

[low shard]

lemme guess, old youtube tutorial

that’s an old original wokada

[low shard]

yeah and it’s bad

[formal ermine]

m

[formal ermine]

a year ago dfhfjksddfhjksadg

[low shard]

uninstall that along with everything you downlaoded off youtube

[low shard]

yeah it’s very old

a year is prehistorica in ai field

[primal nimbus]

and the new wokada looks ?

[low shard]

ai progresses always

-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 si better

[formal ermine]

so ill just download the new one and that'll solve it?

[low shard]

yes

read the new wokada deiteris fork guide

[knotty moth]

there seems less likely anything to fit on less than 8 GB vram, so if you think it's not capable of that, consider some cloud/rental solutions

[low shard]

uninstall all the things you got before

[formal ermine]

ok thanks a lot

[low shard]

yeah, this is an open source ai project, it isn’t chatgpt

[low shard]

yw and lmk

[sharp monolith]

Yeah I'm running on 2GB of VRAM I might have a friend who has an A1000

[low shard]

2gb vram is cooked

8gb vram is barely considered the minimum nowdays

[knotty moth]

that seems like in the era of GTX 460 and some radeon HD

[low shard]

ofcourse the A100 would be better

[woven crow]

@low shard Thank you for mentioning the fork, I hadn’t heard of it yet, I’ve been using the original.

[knotty moth]

note that colab has T4 (15 GB) for free plan

[low shard]

the wokada deiteris fork improved A LOT the performance, and helps with quality because it offers force fp32 (inference) mode

you’re welcome and lmk

[woven crow]

will do 👍

[tranquil radish]

Wdym with should😭

[livid atlas]

Hey guys, I am looking for a way to edit .json files with an llm that I can install locally, can you point me into the right direction? I am very new to AI so every bit of advice would be appreciated 🙏

[simple ore]

1. have a problem 2) decide to use LLM to solve it 3) have two problems

[livid atlas]

So you are saying that it is not a good idea? 😅 I found the results from chat gpt very helpful and I was wondering if I can get that to run locally

[simple ore]

remind me how big is the chatgpt's model ? 🙂

you can run a relatively small quantized models locally, they are not good at predicting large texts

[livid atlas]

Should I be more specific about my project?

I dont want to annoy anyone with my nerd projects xD

[simple ore]

python is great at manipulating json data, should not be too hard to program what you actually want to edit

[livid atlas]

Yeh the editing part is not that hard, but I really suck at that, so I resorted to chatgpt and was surprised how efficient and also accurate it was. I am trying to add a list of Item IDs to an already existing loot pool on my minecraft server

[brittle wing]

does microsoft word has an embed AI?

have*

because editing and aligning is so tedious manually in the word

[olive plover]

What's the point of this?

Also, do I put 0.15 instead of 0.1?

[olive plover]

How?

[olive plover]

Like what? For the Arabic dialect

[livid atlas]

I have heard about something like text to json in this regard, maybe that describes it a little better

[simple ore]

json is a dictionary

key=value, values can be dictionaries

[strange mountain]

so what is the latest version? 💀

[formal ermine]

ok so i downloaded the new ver already the res is still kinda spiking whenever i try to go into a game

[livid atlas]

What are you trying to say

[simple ore]

it is easy to edit

[livid atlas]

I can manually edit them yes but I suck at that and it took me a long time to do, so I tried to make it faster, I also have a large amount of files coming up, so I wondered if I could automate the process, just drop the files, tell the LLM the changes I want to have made and have it done

[stiff goblet]

On Kaggle, I train a voice using Applio. The train time of an epoch seems to be quite long compared to the RVC Mainline fork. Is this because of the vocoders? For example, an epoch takes 15 seconds in Mainline RVC, while in Applio it takes about 40 seconds. It is annoying that it is so slow despite T4X2.

[simple ore]

both on kaggle?

[stiff goblet]

Yes, both on Kaggle.

[simple ore]

mainline trains in fp16, applio in fp32

but it should not be that much slower

[stiff goblet]

Oh, didn't know about that. Is there much difference?

[stiff goblet]

Applio is two times slower than Mainline RVC

[simple ore]

0 chance of model exploding with fp32, with fp16 it may or may not happen.. personally half of my attempts blew up

[stiff goblet]

You mean it's waste of time then ? Maybe i can try to reach epoch 100 to test the model quality..

[simple ore]

fp16 is less stable

[stiff goblet]

Thanks. My dataset length is nearly 20 minutes. I guess that's why it takes so much time.

[simple ore]

there are also some things different, but it should not be 2x slower

i can actually test, I guess

just not on kaggle

[stiff goblet]

Yes, it would be very good to test it. I'm sure I'm using the same settings for Mainline and Applio.

[low shard]

put game graphics to the minimum, tell the game name, close other programs in the background and show a screenshot of the wokada

[low shard]

the wokada deiteris fork b2332 #✨│ai-help message, you have it already dw

[low shard]

better stability and quality at the cost of the delay

[low shard]

lower value = deeper 'masculine' voice
higher value = higher 'feminine' voice
just play with the value till it sounds good

[low shard]

yeah find a better model, not all models are good quality

[simple ore]

so no, mainline is not faster all things considered

[olive plover]

Its 14

[olive plover]

It's a paid model that I have

[olive plover]

Do I put it on and the quality will be better, but it will be delayed?

And Also, do I put 0.15 instead of 0.1?

[low shard]

when did you buy the model and from who?

[olive plover]

A year and a half ago

[olive plover]

Mrm0z

[low shard]

ohh, was just asking since paid commissions are banned now

[low shard]

well yeah because it will need more power

[low shard]

there isn't a right value for everyone, it depends on your own voice and the voice model pitch

[olive plover]

I have 4070
i7-14700

[olive plover]

.?

[olive plover]

Are there any other improvements I should make that improve the sound quality and make it real?

[low shard]

0.1 for fastest voice, 0.15 for improved quality but increases delay by 50 ms

[low shard]

yeah that's good, still means it will need more power and have a bit more delay

[olive plover]

Do I put it 0-15?

[low shard]

your choice if you want better quality

[olive plover]

Okay, the summary of this speech is on put 0-15 so that it becomes a better quality

And the sound will become realistic

[olive plover]

I didn't understand this delay for every 50 ms how

[low shard]

0.15 yes

[low shard]

it uses more power, so more delay

[olive plover]

Okay, is there a huge difference in delay or small?

[olive plover]

And how best in FORK
SIO - REST .?

[low shard]

50ms

[low shard]

what?

[keen crescent]

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[keen crescent]

-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

[grim bay]

Why does only UVR-Deecho-Dereverb works? The rest models in Dereverb tab are just giving error

[olive plover]

How best

sio or rest?

[viscid moss]

Sorry, it will be fixed

[low shard]

leave it at rest

[torpid oriole]

hey someone can you give me linux tutorial how to run that if i double click MMvcServerSIO
the vcclient.log says 2025-05-10 13:24:03,148 INFO [main] Python: 3.10.15 (main, Sep 9 2024, 03:02:42) [GCC 9.4.0]
2025-05-10 13:24:03,148 INFO [main] Voice changer version: b2332 NVIDIA-CUDA
2025-05-10 13:24:03,148 INFO [WeightDownloader] Loading weights.
2025-05-10 13:24:03,168 INFO [Downloader] Verified pretrain/crepe_onnx_full.onnx
2025-05-10 13:24:03,168 INFO [Downloader] Verified pretrain/crepe_onnx_tiny.onnx
2025-05-10 13:24:03,181 INFO [Downloader] Verified pretrain/crepe_full.pth
2025-05-10 13:24:03,182 INFO [Downloader] Verified pretrain/crepe_tiny.pth
2025-05-10 13:24:03,238 INFO [Downloader] Verified pretrain/content_vec_500.onnx
2025-05-10 13:24:03,270 INFO [Downloader] Verified pretrain/rmvpe.pt
2025-05-10 13:24:03,337 INFO [Downloader] Verified pretrain/rmvpe.onnx
2025-05-10 13:24:03,346 INFO [Downloader] Verified pretrain/fcpe.pt
2025-05-10 13:24:03,354 INFO [Downloader] Verified pretrain/fcpe.onnx
2025-05-10 13:24:03,354 INFO [WeightDownloader] All weights are loaded!
2025-05-10 13:24:03,355 INFO [main] protocol: HTTP

[olive plover]

is best sio?

[olive plover]

What is the use of this?

How much should I put it in to improve the sound?

[simple ore]

kinda pitch shift

[olive plover]

How much should be put

[simple ore]

experiment and see what you get

[olive plover]

Force FP32 mode: on (THIS IS OFF BY DEFAULT!) Turning this on improves stability, significantly reduces glitching/artifacting, increases VRAM usage by 200 MB.

I put it on will it improve the sound quality?

[viral mason]

it works pretty well tbh

[low shard]

speak only english here

[olive plover]

How much should I put it in to improve the sound?

[low shard]

nobody knows

#✨│ai-help message

[olive plover]

I will show you the settings that I modified, okay my friend?

[low shard]

yeah they are okay

[keen crescent]

-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

[olive plover]

Force FP32 mode: on (THIS IS OFF BY DEFAULT!) Turning this on improves stability, significantly reduces glitching/artifacting, increases VRAM usage by 200 MB.

I put it on will it improve the sound quality?

[pallid rampart]

Guys, please help! What should I choose here so that it works for me?

[low shard]

yes they are fine

[olive plover]

sure?

What about the CHUNK

[low shard]

either use client (can use echo and noise suppression) or use server with wasapi/ASIO for less delay https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

close everything in the background

are you using it with a game?

[low shard]

if you want you can lower it a bit for less delay but just remember it to be higher than perf value by a little

[pallid rampart]

No! How do I set these settings here? I just forgot

[low shard]

what program are you using the wokada deiteris fork with?

[pallid rampart]

:_)?

[olive plover]

How much should I put please?

[low shard]

it doesn't need to be precise

just a bit lower if u want less delay

[low shard]

?

[olive plover]

okay thx

[pallid rampart]

Just tell me please where I need to select the microphone in this program. in input or output? and where I need to select the virtual cable here

[low shard]

there's 2 different settings

client (more delay but easier and can use echo and noise suppression:

• input: microphone
• output: line 1
• monitor: headphones

or you can use server with wasapi/ASIO which is harder and can't use echo/noise suppression but helps alot with delay https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

it's your choice

also

you seem lagging alot in the pic

which is why i'm asking what program are you using the wokada with

[pallid rampart]

I have windows 11

[low shard]

you need to choose either if to use server or client

what do you need?

they provide different things

as i explained you

and yes you seem lagging

[pallid rampart]

i need to use voice changer

[low shard]

yes.. and i explained you the 2 different audio setups with their perks #✨│ai-help message

you need to choose what fits you

i explained you the pros and cons of each

the program isn't as easy as like chatgpt it's open source ai

[pallid rampart]

thanks

[viral mason]

https://cdn.discordapp.com/attachments/1365777183804756021/1369148522716336220/IMG_1886.gif

[unkempt furnace]

cpu
No eligible .pth files found in ./models\volkan_konak_model2 bu I have pth file can anyone help me

rvc gui

[low shard]

rvc gui outdated asf

don't use video tutorials for rvc

what's ur pc gpu and what u want to do

[unkempt furnace]

rtx 4070 mobile I do an rvc model but I cant use that

can you help me

@low shard

[low shard]

do you want to do inference or training

[unkempt furnace]

inference I guess

I make a sound I havep pth folder already

[quaint dock]

Hey there, I'd like to ask how I could "pretrain" a voice dataset (like just the raw voice itself) I came across a model which said either "pretrained with Hubert" or with Titan.

[unkempt furnace]

hey I came first

[formal ermine]

min graphics, tried it on roblox for now, no programs, i couldn't send a screenie rn cuz perms (slr)

[low shard]

inference = use model
training = make models

[unkempt furnace]

okey inference

what will I use

[low shard]

As you got a good PC, you can use RVC locally, you can choose between:

• Applio: A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
• Mainline: The original RVC

[low shard]

!give-media-perms 1h @formal ermine

[formal ermine]

here

[low shard]

pretrains are just the base used for training other models

[formal ermine]

I closed the game already so res is normal and i haent eevn started it

[low shard]

use rmvpe without onnx, show a screenshot while in game

[formal ermine]

tried while in game

[low shard]

use chrome or firefox, might help since operagx is known by other users to give issue especially for its fancy effects eating more resources

lower extra to 0.4

be sure roblox settings are to the minimum

[unkempt furnace]

how can I use applio

[formal ermine]

much better thanksss so much

[low shard]

read the guide

[low shard]

yw

[lethal dirge]

@still crystal

[analog obsidian]

this setting is meant to decrease the amount of robotic breathing and sibilants
at a value of 0.5 this is disabled, anything lower than 0.5 enables it

original rvc devs used a value of 0.33 by default for inference

i havent noticed a big difference of 0.5 vs 0.33 to be honest

they sound exactly the same

is supposed to do something but ehh, a good model have natural sibilants and breathing by default

yea it only affects index accuracy

they're slightly better using 0.33

at 0.5 they artifacted a bit more

[hot depot]

what's the recommend amount of epochs for an hour of audio? or is an hour of audio just overkill?

[crude flame]

There is no way to know how many epochs is enough. You will have to use the tensor board and listen to the epochs to find the best epoch. An hour of audio is fine but anything over doenst make a difference.

[hot depot]

I can lets say, do 500 epochs of training, save the modle, and then do the other 500 epochs tommorow though right?

so I don't have to do it all in one shot

[crude flame]

you can save an resume yea

[hot depot]

alr let me cook up

[analog obsidian]

using index and protect voiceless consonants seems to improve breathing a bit

0:07 breathing is better at protec 0.1 and index 0.75

f0 used for training too, fcpe has different breathing compared to rmvpe
but at the end robotic breathing/sibilants occur randomly, so while protect voiceless help, eventually you'll get some robotic breathing/sibilants due to limitations in the tech

[devout field]

hey guys is there an ai that generates music based of off other music? Im tryna change genres of a song

[river gulch]

That explains why the first model I train there was kinda off compared to RVC

Not relative with it, but I gonna try to use RVC then, even if it ran exaggerate slow on my laptop, but even with 50 epochs, it did generate better ones.

Sorry to but in.

Simple question: What is better for Text to Speech? Applio or Free User Elevenlab?

[still crystal]

?

[austere siren]

how come the voice changer does not work on roblox?

i changed my input and everything

[solar thunder]

hii can someone help me i dont know how to dowload it where to even start from

[fiery rain]

virtual audio cable not workin

[inner kayak]

okay weird question, but how would i go abut making a voice changer ai tof my friends voice? like i want to be able to speak in their voice. (i do have written permission to do so)

[brittle wing]

Hey, does anyone of you use
deevid
krea
pollo.ai
???

[craggy bough]

stop spamming

[brittle wing]

I'm just asking, does anyone use the sites???

[craggy bough]

so you should ask it once and wait for a reply, not spam it across 4 channels

[brittle wing]

doesn't matter, the main thing is that I get my answer

[robust condor]

i just switched over to the fork web version of okada after using the old one for so long i closed the software and to open it again do i just open MMVCServerSIO?

[grim perch]

when uploading a model, how do i add a photo?

[hot depot]

I'm on epoch 60 rn and I totally forgot about your message, can I still install this tensor board thingy while training the model?

[crude flame]

yup

[hot depot]

do I just look it up and download it or is there something special I have to do in rvc?

[crude flame]

do you have applio?

[hot depot]

no

idk what that is

this is my first time training a model

[crude flame]

read this https://docs.aihub.gg/rvc/resources/training/#tensorboard

[hot depot]

its says that my rvc folder can't contain any spaces in the file path, however mine does, and I can't changing it because it says the folder is in use. Should I just ignore it and move on?

[lyric geode]

Hey y'all the MP3 download for my song cover isn't as high quality as when I listen to it & record it myself. Do I have to purchase weights to get the highest quality version?

If so I was hoping someone who does pay for weights could just shoot me a download for it because I'm trying to make an AMV with it

[hot depot]

sorry to keep pinging you, but I seemingly have tenserboard fully installed, however it doesn't open anything on my browser and when I copy the address it just opens to a white page. Any advice on how to fix?

[crude flame]

Just copy the local host part

Not the whole link

[hot depot]

I'm fairly sure I'm doing that. Could I have something in my browser or internet setting misconfigured?

[crude flame]

Try a different browser

[crude flame]

Let's keep support issues here and not dms so if others have this issue they can find the answer here

Tbh I've never seen that issue before

Or I just don't remember

[hot depot]

if I inspect element if tells me this

so wtf is mime type checking

[crude flame]

Did you happen to close the cmd line

[hot depot]

no

[crude flame]

Ok good

You tried refreshing right?

[hot depot]

yes

[grim perch]

I have 20 min of clean data set audio. How many epochs should i train the model to. Its a low male speaking model

[solemn grail]

How feasible would it be to train a model to clean up manga pages after removing the text? Or are there any already?

[quaint drift]

how can I put the metadata.json in voicechanger mmcv

[misty haven]

hello how do i use the voice

[craggy bough]

!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.

[misty haven]

Hello, I would like to know how to use the voice changer, what app to download, how to set it up, etc. Thanks

is this better dawg

[placid salmon]

did you ever find a solution to this?

im having the exact issue

[modern swallow]

kaggle only allows 40 minutes of use right?

[modern swallow]

Will creating a new account help me continue using it?

[knotty moth]

you can't save and transfer the whole session state
for rvc training, download the whole model's logs folder in order to resume later in anywhere

[brittle wing]

Is de-reverb & deecho by sucial a good model option?

[night bay]

im trying to use VoiceAi and im stuck with payment section. I head it this is free app.

[knotty moth]

they want to grab more money from you, while there's open source alternative

[night bay]

oh, is thee any other app that i can use?

[knotty moth]

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

[fathom ore]

could anyone here pleaseeee help me with male/female vocal seperation? 🙏

[untold hull]

Hey everyone! 👋

I’ve recently seen some examples of Gemini Flash 2.0 Experimental being used for image generation inside tools like ComfyUI, possibly through custom nodes or API integrations.
Does anyone know if there’s an official API, beta access, or any documentation on how to set this up?
Or maybe how to connect it via Vertex AI as part of a programmatic workflow?

Would really appreciate any insights or experiences you can share. 🙏
Thanks in advance!

[misty haven]

MMVCServerSIO how do i open this

[low shard]

Elaborate

• what's ur PC GPU
• what u want to do
• what's the issue

[misty haven]

they say open it in the guide

and i cant

[low shard]

Elaborate what I asked u

[misty haven]

i want to use the voice changer deiteris and MMVCServerSIO doiesnt open i cant send ss here

[low shard]

!give-media-perms 1h @misty haven

Now u can

[low shard]

This is only for realtime btw, like for calls not pre-recorded audios

Also what's ur PC GPU

[misty haven]

yea ik

[misty haven]

3080

[low shard]

-realtime1

-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]

Did u read the 1st link guide?

[misty haven]

which ones the realtime then

i got this downloaded it doesnt change my voice

idk why

[low shard]

Don't follow video tutorials

Those are old
That's original womada

Wokada*

-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 1st link

Wokada deiteris fork

[misty haven]

i did bro

[low shard]

That's not where the guide directs you at all

[misty haven]

ok what do i do then

[low shard]

You have to read it and get the Nvidia windows download

Forget everything off youtube

[misty haven]

sec

[low shard]

?