✨│ai-help

[pastel oak]

Deiteris for amd best
For nvidia theres a new version of the og rvc that seems to be running slightly better, staff team is working on providing a prepackage with a guide soon enough. Until then deiteris still best paid and unpaid

[outer valley]

ohhh? I have an 4090 so that sounds interesting

[outer valley]

Deiteris is really good. Just needs a good model to work with.
I still hope like AI art with hands... the pitch and emotional stuff get better with RVC soon 🙏

[patent loom]

quick question what if ther`s no index file??

[pastel oak]

I higjly doubt there will be much of a difference for you then

[pastel oak]

Dont need index file its optional for all use cases

[patent loom]

alright thx

[outer valley]

hmmm i see what you mean.
Still thanks for the info.
About the general question with pitch (going out of range) is there hope for big steps for things like that this year?

[pastel oak]

Impossible to tell for now

At least with my current knowledge lol

[outer valley]

I see.
Is there anyone here who might be able to answer this question?
It seems that RVC is stagnating. Or am I just mistaken?

[fluid topaz]

It is stagnating, there isnt any big improvement since the last 2 years

[winged ember]

hi, does anyone know when free training voice on weights.gg what are the parameters for it ? like how many steps and stuff?

[unreal linden]

Hello, I get the high ping right now, and I'm currently running gpu for some project. Is it the reason?

[unreal yarrow]

Hello sorry to ask here but there is nothing specified in the channel list or if it is I am stupid and apologize. Question is , is there a moderator here or someone I can talk to about an issue?

[hallow thistle]

Is there something wrong in this server? Or just RVC/W-Okada?

[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.

[unreal linden]

Hain?

[hallow thistle]

Without mentioning either W-Okada or realtime voice change, I have no idea what you're looking for. So please elaborate more about the issue you encountered.

[unreal linden]

I mean, I'm saying that I'm training huge dataset for my project and when i open realtime voice changer , ping is very high around 5k

[hot lagoon]

Which one do you guys put for more?
Leaving the original backing vocals
Or converting them

[unreal linden]

And I downloaded from deiters realtime voice changer link

[knotty moth]

so what's the point?

showing that you are training a model while running voice changer using a 4090?

[radiant ledge]

hello gus im new to this just wanted to know is it possible to use these voice models for like tts (text to speech ) if so how would i go about making it happed just needed it for a project

[urban forum]

Strange.

Did they purge channels in here?

[hallow thistle]

Help channels for W-Okada and AI image generatoer are either archived or deleted. This is formerly RVC help channel that rebranded into #✨│ai-help.

[urban forum]

Oh.

So there's no more support for RVC?

[analog obsidian]

help channels were merged to one channel (this one)

so you can ask rvc, wokada, etc questions here

[brittle wing]

can anyone help me with one thing?
i want to train ai, and preferably build a software which:

• can give me clips based on my preferences from youtube videos where i will provide clips to train ai ,
  -and the ai generates clips which but are unused and have potential to go viral.
  -Here's an example

[whole crypt]

when i merge voice models in w-okada, the resulting merge file won't load properly. i get a bunch of errors and the console says AttributeError: 'RVCInferencerv2Nono' object has no attribute 'use_jit'

what does that mean??

i think it also converts the .pth to .safetensor in the process. what's with that?

[atomic topaz]

cpu_fallback

Voice changer beta using CPU instead of GPU
how to fix it

[hallow thistle]

Which W-Okada are you using? And what is your PC GPU?

[atomic topaz]

vcclient_win_std_2.0.76-beta.zip
rtx 2070 maxq

[hallow thistle]

Topics about RVC, W-Okada, AI image and many others all merged into #✨│ai-help and #1192011222023950368.

[hallow thistle]

Download and use this better W-Okada from this guide instead. https://rentry.co/ForkVoiceChangerGuide#download-for-nvidia-gpu-on-windows

[atomic topaz]

alright thanks

[mighty flint]

i'm also trying w-okada with amd rx 7900xtx, on wsl2 linux. It doesn't recognize the gpu

[hallow thistle]

Did you follow this guide? https://rentry.co/ForkVoiceChangerGuide#download-for-amd-gpu-on-linux

[mighty flint]

yes

[hallow thistle]

Never used Linux. But make sure you have updated your AMD Radeon driver for Linux.

[whole crypt]

i'm using the w-okada fork for AMD and for whatever reason, when i make a merged voice model, it causes errors when i try to load it. normal models load just fine. it also converts them to safetensors. any idea why this is happening? someone please help. the error code was AttributeError: 'RVCInferencerv2Nono' object has no attribute 'use_jit'

[little igloo]

can anyone help me train a model on yolo 8 using google colab, i have everything i need but cant start training, i keep getting errors

[white ice]

I need help please, my gpu doesnt work on w okadas deiteris voice changer it displays an error and doesnt work during the voice conversion

how can i fix this

[fleet shadow]

which is the best best app to make a voice model ?

[odd shale]

Applio.

There you have a guide about how to learn to make voice models.

https://docs.aihub.gg/essentials/how-to-make-voice-models/

[sly furnace]

heyy eddy!

[viscid moss]

hewwo

[sly furnace]

so how much duration

let me first get things ready

[viscid moss]

I think 30 mins it's fine

[sly furnace]

and which app should i use to record?

[sly furnace]

anything more benifits ? i dont need artifacts 😭

i dont want people to detect the ai

[viscid moss]

any

[viscid moss]

maybe 1 hour then

[astral river]

Peace be upon you, can you convert this clip to Hussein Al-Jasmi's voice using RVC?

I recorded it with my voice and with the same melody, and I want the result to be the same performance but with his voice.

Thanks in advance!

[sly furnace]

okay ill msg you after i am done

[viscid moss]

oh, lyery

[analog obsidian]

thats impossible, you can make the model sound "natural" but eventually they will find out its ai

rvc doesnt sound human at all

[sly furnace]

i dont think so , i have seen many people make ai cover

that sounds next to natural

[viscid moss]

they can be more helpful than me

:p

[sly furnace]

i am learning from everyone , you are mod for a reason 😭

[sly furnace]

what should i do then sir ?

[viscid moss]

ye cause i like to ban ppl

[sly furnace]

😭

but u my friend 🤞

@analog obsidian can you help sir ?

[brittle wing]

pls anyone 🥺

https://tenor.com/view/sullimvs-jinrinotes-jinnielabs-dog-crying-gif-22771721

[idle moon]

y'all be using auto tune ?

[hidden zenith]

guys can the replay app conflict with own ethernet connection?

[low shard]

elaborate

[white ice]

Once I put my gpu option, it gives me an error

And once I start voice conversion, error pops up on the upper right side constantly on click

[maiden raft]

hey guys im coming back after over a year, is applio still the best?

[broken urchin]

well i dont know if its the best, but i used Applio and made 2 voice models and they work great for me so

[maiden raft]

well I don't expect for it to be broken but i'm asking if there are alternatives which give better results

[broken urchin]

i dont know lol

[wheat copper]

why does the website for rvc not load after i put in the code

[low shard]

what's your pc gpu? what version of wokada deiteris fork did you use? also show a screenshot of the issue

!give-media-perms 1h @white ice

[low shard]

yes, it still gets updated

[low shard]

it's not an RVC website, you're using google colab, a cloud (remote good pc) computing provider that runs code on good hardware

share the google colab link you're using, along with your pc gpu name and also what you want to do

[wheat copper]

https://colab.research.google.com/github/hinabl/RVC-Online/blob/main/Mainline_Colab_Full.ipynb?authuser=2#scrollTo=CKn1Ha9OH_pU

[simple ore]

outdated colab

[low shard]

outdated

[simple ore]

it is dead

[wheat copper]

an i have a 7900xtx amd

[low shard]

** always check #📰│dev-updates **

mainline colab is broken since months

[low shard]

why are you using cloud then

you can do it locally (runs on your pc hardware)

[wheat copper]

i dont know how to train locally with amd

[simple ore]

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

do not skip steps, read, understand, so as it says

[low shard]

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

[wheat copper]

everywhere ive looked it says it doesnt have amd options

[low shard]

lmao same timing

[wheat copper]

oh

is applio better or on par with mainline rvc?

[low shard]

it does with Zluda (CUDA emulator) for AMD

[simple ore]

if you want to use your fancy 7900 it is

[low shard]

Applio is an RVC fork, a modified version that gets updated

it's easier to use + has performance improvements

[wheat copper]

oh yeah i forgot

sorry i havent done this stuff in awhile

[wheat copper]

wwouldnt it take super long since im using amd and it needs to, well emulate and all? or is my gpu good enough to negate that

[low shard]

it should run fine

@simple ore used it before getting an nvidia gpu

[simple ore]

7900xtx is a beast

[wheat copper]

thats nice to hear, then is there a way to also get it to be used by okada?

my cpu handles okada for some reason even though i think i set up everything right

[simple ore]

use the voice changer fork for amd

[wheat copper]

mind telling me where to find it?

[simple ore]

https://rentry.co/forkvoicechangerguide#download-for-amd-intel-and-cpu-on-windows

[gilded crown]

yooo guys what RVC processing gives better results - crepe or harvest? is there any better?

[wheat copper]

this is quite confusing, i have it but after that im getting errors when trying to use it

[simple ore]

both of those methods are outdated af

[simple ore]

how did you get this error?

[wheat copper]

im not sure, i downloaded the amd one then changed cpu to my gpu, added a model, then that came up

[simple ore]

so in GPu you select 7900xtx directml?

[gilded crown]

damn i'm using RVC GUI so what i wanna do to get the best results ToT

[wheat copper]

yes

[simple ore]

full log then

[wheat copper]

huh?

how do i full log?

[simple ore]

just copy the whole text

[low shard]

RVC GUI outdated asf

don't follow yt tuts for RVC

A year old tutorial is extremely old in ai field

@gilded crown what's ur pc gpu and what do u want to do

[wheat copper]

it randomly starte working wwhen i restarted, nevermind!

[low shard]

rebooting fixes everything 🙏

[wheat copper]

just wondering, could you help me with the installation of applio or no?

sorry if all the asking for help is a bother

[simple ore]

download the compiled version, unzip to C:\Applio or D:\Applio

https://huggingface.co/IAHispano/Applio/blob/main/Compiled/Windows/ApplioV3.2.9.zip

then follow the guide above step by step, dont skip anything

for 7900xtx use hip sdk 6.1.2

[modest willow]

having a lot of trouble with rvc since upgrading my gpu

2025-04-30 14:12:08,338 INFO [main] The server is listening on http://127.0.0.1:18888/ 2025-04-30 14:12:08,338 INFO [main] -------- 2025-04-30 14:12:14,678 INFO [MMVC_Namespace] Connected SID: QZhRzc84MaYqe8KIAAAB 2025-04-30 14:12:39,583 INFO [VoiceChangerManager] update configuration gpu: 0 2025-04-30 14:12:39,584 INFO [DeviceManager] Switched to 0: NVIDIA GeForce RTX 5070 Ti (CUDA) (cuda:0). FP16 support: True 2025-04-30 14:12:39,717 ERROR [MMVC_Rest_Fileuploader] CUDA error: no kernel image is available for execution on the device CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1 Compile with TORCH_USE_CUDA_DSAto enable device-side assertions. Traceback (most recent call last): File "restapi\MMVC_Rest_Fileuploader.py", line 48, in post_update_settings File "voice_changer\VoiceChangerManager.py", line 199, in update_settings self.vc.update_settings(key, val, old_value) File "voice_changer\VoiceChangerV2.py", line 80, in update_settings self._generate_strength() File "voice_changer\VoiceChangerV2.py", line 102, in _generate_strength * torch.linspace( ^^^^^^^^^^^^^^^ RuntimeError: CUDA error: no kernel image is available for execution on the device CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1 Compile withTORCH_USE_CUDA_DSA` to enable device-side assertions.

2025-04-30 14:12:45,049 INFO [MMVC_Namespace] Disconnected SID: QZhRzc84MaYqe8KIAAAB
2025-04-30 14:12:46,691 INFO [MMVC_Namespace] Connected SID: i6zo-M1CDfg_ekg6AAAD`

whether it's applio, the modified w-okada in one of the guides here, mangio-rvc or the original rvc-project. i dunno how to fix it

[low shard]

that's because the 50 serie is pretty new and not as much supported

[modest willow]

uwah

[low shard]

How to (unofficially) use Applio for RTX 50 serie cards

Follow to download it as said it in https://docs.aihub.gg/rvc/local/applio/

After you extracted the precompiled, go to the path in Windows explorer, write "CMD" and press enter, then in CMD write env\python -m pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --upgrade --index-url https://download.pytorch.org/whl/cu128

If you get any already satisfied requirement issue, run env\python -m pip uninstall torch torchvision torchaudio then the command said above

for wokada, just use the wokada deiteris fork which is better anyways and has 50 serie support https://rentry.co/forkvoicechangerguide#download-for-nvidia-rtx-5000-series-gpu-on-windows

those methods should work without any issues

@modest willow let me know how it goes ofc

[modest willow]

https://tenor.com/view/i-kneel-bejita-vegeta-dragon-ball-z-dbs-gif-9873872880292680534

thanks king

[low shard]

also, mangio rvc is abandoned since 2023, and original rvc (aka mainline) isn't as much suggested as Applio in terms of performance and ease to use

wokada deiteris fork (which I sent) is also way better in terms of performance compared to original wokada

[modest willow]

this is very good info to have, i appreciate it

[low shard]

yw, be sure to update your drivers (latest nvidia driver was today), everytime i check them there's a shit ton of 50 serie fixes so it could help ya out

[modest willow]

can models be made with the okada fork or is it strictly for voice inference? and lmao i'll download it right now

[low shard]

wokada deiteris fork is strictly just made for realtime rvc inference for calls

for model training, use Applio

[modest willow]

yeah i figured

big mcthankies

[low shard]

the whole channel is just 50 serie fixes 😭

[low shard]

you're welcome, let me know

[simple ore]

that command is outdated

for applio 3.2.9 the command is env\python -m pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --upgrade --index-url https://download.pytorch.org/whl/cu128

[simple ore]

if you have questions, feel free to ask

[gilded crown]

welp i knew the GUI for a long time and i'm on a new pc now so wanted to download smth to edit my voice

its just for a regular meme-like voiceover

to record myself and transform it with a RVC model

like 5s long stuff

what are other options than rvc gui

just for personal use

[low shard]

oops, thx

[low shard]

alr it seems pre-recorded audio inference (use models)

what's your pc gpu?

[gilded crown]

intel iris xe

like integrated

[simple ore]

click initialize, the actual error is in the other window

but it seems you did not add a model

[wheat copper]

it didnt give me a install bat thing

wait nevermind i think i di something wrong

[white ice]

My gpu is GT640 Nvidia

[white ice]

@low shard

[modest willow]

it works \o/

[white ice]

And I use AMD Intel version

[modest willow]

the real time adjustments are actually so convenient though oml

[low shard]

cooked, wouldn't even be supported

it can run but on cpu super slow

it's better you use cloud

Train (make) RVC Models on cloud:

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

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio (UI)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • 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

[low shard]

then you have your answer

that's not even supported by CUDA, which is a fundamental thing for AI

the gpu is basically prehistorical and wayyyyy too weak for the program to even run

you could try the cpu mode but it's ofcourse unstable since a cpu is not made for such heavy tasks

[gilded crown]

welp okay thx gonna check that out

[simple ore]

you likely did not unzip it properly.. use 7-zip, winzip, winrar or whatever, not the default windows extraction

[low shard]

@white ice the only way you can use wokada deiteris fork is by cloud (remote good pc), since your pc is extremely too weak to run it locally

the only working cloud is:
W-Okada's Deiteris' Fork Voice Changer Kaggle, which gives 30 hours weekly of free gpu, but you need to login and also verify your phone number, also be aware that cloud gets updated by their provider so ofc it's not as stable as local

[white ice]

Thank you

[low shard]

yw and lmk

[wheat copper]

i got it

now its just eating up my cpu

[simple ore]

what's in the console window?

[wheat copper]

it said it was using cpu

i just stopped it

im just gonna restart the whole installation process

[simple ore]

dont skip steps in the guide

[wheat copper]

i think i did lol

sorry for wasting ur time there

😅

[raven ginkgo]

ik this is pretty old but u can just use a noise gate plugin for example in audacity to remove any quiet noises

[wheat copper]

i followed the steps and i have it up but it doesnt use my cpu

[whole crypt]

please someone at least respond, i've asked three times now... i'm using the w-okada fork for AMD and for whatever reason, when i make a merged voice model, it causes errors when i try to load it. normal models load just fine. it also converts them to safetensors. any idea why this is happening? someone please help. the error code was AttributeError: 'RVCInferencerv2Nono' object has no attribute 'use_jit'

[simple ore]

then you missed something

show me this

[simple ore]

okay, what's in Applio folder?

[simple ore]

what the f is that

you need to install HIP SDK into its default folder

program files/amd and so on

[wheat copper]

oops

[simple ore]

and dont forget to add ROCM/bin path to the environment variable's Path

[nocturne patrol]

i put input as line 1 right

ohh

[low shard]

in other programs, yes

[wheat copper]

well i think it works now, its just stuck on this when i wanna train

is that normal

[nocturne patrol]

i cant hear myself in discord mic test

[simple ore]

you really should've read the instructions, as I said

[wheat copper]

should i just start over

[wheat copper]

oops

sorry!

it did something like that before so i thought that was it

[low shard]

share a screenshot of ur wokada and discord settings

[simple ore]

training has more tasks to compile

[nocturne patrol]

alr

ignore the picture it wont let me change the names so im doing it based on picture

[simple ore]

you need to send output to a virtual cable

and set it as input in discord

[nocturne patrol]

oh

are u sure

[simple ore]

and send discord output to headphones

otherwise it will be picked by mic

and go back to voice change and to discord as input

[ripe mantle]

can i use the voice changer into roblox voicechat or no?

[nocturne patrol]

i dont know what ur saying

put what as input

i will just wait for nick thanks for trying

[low shard]

wokada: output: line 1

also u can uninstall vb audio cable

[nocturne patrol]

alr

do the rest stay the same

[low shard]

yes

[nocturne patrol]

ok thx

[low shard]

yw

[crystal gull]

you can use it with anything that takes in standard mic input

which would be.. roblox

afaik roblox shouldn't block it

[steady folio]

why does my http bat file close instantly as i try opening it?

[whole crypt]

please!!!!! someone at least respond, even if it's barely any help, i've asked FOUR times now... i'm using the w-okada fork for AMD and for whatever reason, when i make a merged voice model, it causes errors when i try to load it. normal models load just fine. it also converts them to safetensors. any idea why this is happening? someone please help. the error code was AttributeError: 'RVCInferencerv2Nono' object has no attribute 'use_jit'

[whole crypt]

i appreciate the IDK at least 😭

[viscid moss]

It seems you are the first person to have that error

[whole crypt]

well if any techhead wants to figure out why its happening my DMs are open 😭

[viscid moss]

Dietieris fork right?

[whole crypt]

it's super confusing. i made a merge that worked TODAY, and then later the same day they started failing

[whole crypt]

yeah

maybe i should just try to reinstall..

[viscid moss]

try

[simple ore]

depending on how you've done the merge, perhaps there's something missing

it may help to see a full error log

but I think there's a bug

[whole crypt]

2025-04-30 22:47:36,211 ERROR    [VoiceChangerManager] 'RVCInferencerv2Nono' object has no attribute 'use_jit'
Traceback (most recent call last):
  File "voice_changer\VoiceChangerManager.py", line 212, in change_voice
    audio, vol, perf = self.vc.on_request(receivedData)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "torch\utils\_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\VoiceChangerV2.py", line 162, in on_request
    result, vol = self.process_audio(audio_in)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\VoiceChangerV2.py", line 126, in process_audio
    audio, vol = self.vcmodel.inference(audio_in)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\RVCr2.py", line 259, in inference
    audio_model = self.pipeline.exec(
                  ^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\pipeline\Pipeline.py", line 235, in exec
    out_audio = self.inferencer.infer(feats, p_len, pitch, pitchf, sid, skip_head, return_length, formant_length).float()
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\inferencer\RVCInferencerv2Nono.py", line 57, in infer
    with torch.jit.optimized_execution(self.use_jit):
                                       ^^^^^^^^^^^^
AttributeError: 'RVCInferencerv2Nono' object has no attribute 'use_jit'```

[simple ore]

dml\MMVCServerSIO_internal\voice_changer\RVC\inferencer\RVCInferencerv2Nono.py

it has with torch.jit.optimized_execution(self.use_jit):

but there's no such class variable defined, there's instead self.use_jit_eager = not use_jit_compile

[whole crypt]

ok i have this file

[simple ore]

so it may work if you change the code to with torch.jit.optimized_execution(self.use_jit_eager):

[whole crypt]

is there any reason why it'd be buggy like that?

[simple ore]

for that we need to ask the author

[analog obsidian]

original w-okada is buggy asf
the fork is an edit of that
so

[whole crypt]

well idk why its a bug only im having, i guess is my point

i feel like that points to something weird in my system?

but i havent tweaked it at all and it still runs fine for normal models. i can merge things fine too. just cant use merged .pth files.

[simple ore]

not original wokada

the fork

[whole crypt]

i'm willing to try this but idk how

[whole crypt]

im using the fork

[analog obsidian]

the fork is also a buggy mess

[simple ore]

I think you're may be trying to run it without f0

so it hits that module instead of f0 where there's no bug

[whole crypt]

ok, what... does that mean? how do i fix it?

[simple ore]

pick f0 extractor

[whole crypt]

is this something i need to download right now? searching f0 in the wokada folder gives me nothing

[whole crypt]

ohh, in the GUI. ok, checking

[knotty moth]

sorry but your attention seeking attitude seems rather annoying

[whole crypt]

...i'm sorry if i upset you? i just want help because i'm out of my depth. i'm not trying to bother anybody.

[analog obsidian]

try changing the f0 and see if it fixes the problem

if it works try rmvpe onnx again

[knotty moth]

what version are you using? the fork voice changer for AMD?

[analog obsidian]

i'd do what noobies said tho

[whole crypt]

it gives the error on every f0 version

[whole crypt]

yes, fork for AMD

[analog obsidian]

try doing what noobies told you

[simple ore]

weird, because I have this version running just fine

so perhaps the model has no pitch guidance, very old

[whole crypt]

could it be something i'm messing up in the merging process?

[analog obsidian]

how old is the model you downloaded?

[simple ore]

depending on how you merged the files, it may not copied the config properties

[whole crypt]

they vary in age, i'm unsure. all i have is the little thing with the sliders, on how much each voice model can be input

[simple ore]

where did you get it

[knotty moth]

you seem to have messed up on JIT compilation in advanced settings
but make sure you're using the latest from the guide https://rentry.co/forkvoicechangerguide

[analog obsidian]

are you able to use both models individually? without merging them

[whole crypt]

yes.

i also just tested making a merge using only one model input, and the output was buggy too.

[simple ore]

so whatever merge application you used, it did not copy 'f0' = true

[knotty moth]

you can try merging models in Applio, then import the merged model to the voice changer

[whole crypt]

i'm using the one in w-okada.

[analog obsidian]

hmm maybe the merging is bugged in the fork

i would merge the models either in applio or mainline

[simple ore]

that's why the voice changer goes to use non-pitch guidance code

[whole crypt]

it worked earlier today for me which is especially odd. but i'll try using applio, i guess

[knotty moth]

afaik the model merge in Applio works well, even for 32k sample rate models

[whole crypt]

should i redownload the entire thing?

[knotty moth]

Latest Version b2332 from December 2024

check the version in your gui

[whole crypt]

you mentioned issues in advanced settings, do you mean here or in a more complex part of the setup?

[simple ore]

changing this would not help

because the code is using a variable that does not exist

[whole crypt]

b2332. i downloaded it earlier this year.

[analog obsidian]

you found a new fork w-okada bug

congrats i guess lol

[whole crypt]

hooray!

[simple ore]

even if you fix the code the merged model without f0 flag wont work properly

because it was trained with f0 and needs pitch guidance enabled

so fix the merge

[analog obsidian]

i don't think he downloaded a 2023 model....
right?

im like the only person training in mainline atm lol

[whole crypt]

i have absolutely no idea when the voice models are from, i've had a few of these for a while. would that cause a problem?

they all work perfectly on their own

[analog obsidian]

try making one of the models sing a song

if it always sound monotone, it wasnt trained using pitch guidance

and that may be causing the bug

well technically not a bug

lets say the issue

[simple ore]

here

[knotty moth]

the models not using pitch extraction may seem too old

[simple ore]

easy check

go to model_dir/x/

whatever the number is

there's .safetensors file

open it in notepad

[knotty moth]

the number is the slot number

[analog obsidian]

i still dont know why rvc boss added such option in his trainer

[simple ore]

or actually look at params.json

i bet f0 value is false

and for the source models you used it is true

[analog obsidian]

TIL the fork has this lolol

[whole crypt]

f0=true in params for working merge i made ages ago, f0=false in fresh merges that do not work. that's it.

does switching that line of text fix it, or is it inherently gunked?

[simple ore]

so you used a shitty merge

[analog obsidian]

rip

[simple ore]

you can try setting it manually to true

[whole crypt]

its the one w-okada gave me

[simple ore]

but safetensors file also has the flag and I dont know which one it would use

[analog obsidian]

applio/mainline merge is better anyway

[simple ore]

safetensors:

[whole crypt]

you're right, it now says Pipeline is not initialized.

after switching params.json to be f0=true

[whole crypt]

i'll try applio

[simple ore]

after restarting the application?

what else does it say in the log?

[whole crypt]

    raise PipelineNotInitializedException()
Exceptions.PipelineNotInitializedException: 'Pipeline is not initialized.'
2025-04-30 23:10:29,080 ERROR    [VoiceChangerManager] 'Pipeline is not initialized.'
Traceback (most recent call last):
  File "voice_changer\VoiceChangerManager.py", line 212, in change_voice
    audio, vol, perf = self.vc.on_request(receivedData)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "torch\utils\_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\VoiceChangerV2.py", line 162, in on_request
    result, vol = self.process_audio(audio_in)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\VoiceChangerV2.py", line 126, in process_audio
    audio, vol = self.vcmodel.inference(audio_in)```

[simple ore]

oh well, use a proper merge then

[whole crypt]

yeah, i'll try applio, hopefully it's just a weird bug. it's just so weird that i was able to create a working merge this morning using this method, and then something suddenly changed and they're all borked for good.

[hushed rover]

hello! i'm new to this and i'm trying to make an AI cover using google collab (this one https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?usp%3Dsharing&sa=D&source=editors&ust=1704303145687891&usg=AOvVaw3M9tmokG80RXF-GD1LJqCL#scrollTo=PNIn-NKDjYJ3 ). is there a new version already or that's the newest, anyone know?

[simple ore]

probably this https://colab.research.google.com/github/Eddycrack864/RVC-AI-Cover-Maker-UI/blob/main/assets/RVCAICoverMakerUI.ipynb

[hushed rover]

thanks, i'll try!

[simple ore]

@viscid moss btw, "!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit" are no longer needed, google finally fixed their shit

[hushed rover]

i tried using this link, already at the last step but i cant run the public url. i got 504 gateway time-out. anyone know to fix this?

[hushed rover]

I used this link https://colab.research.google.com/github/Eddycrack864/RVC-AI-Cover-Maker-UI/blob/main/assets/RVCAICoverMakerUI.ipynb#scrollTo=sDo4MkcLcN5l
and i've tried to start UI using gradio and localtunnel but both doesnt work.
the local tunnel one didnt generate public url after 10 minutes (i already tried it thrice).
the gradio one generated public url, but it's not running. i always get 504 gateway time-out (i also tried it thrice)
do anyone know how do i fix this? thankyou in advance!

[hallow thistle]

Most of RVC Colab notebooks are broken at the moment.

[long oyster]

hello, I just scanned my computer and accidentally deleted the MMC SERVERSIO.EXE app. can someone help me?

[stiff knoll]

hello @hallow thistle

i want to write a text but the model can not undersatnd the words amount that i want

i am greek and by english is not the best excuse me

[hallow thistle]

If one of your antivirus programs removed the program, please make sure to allow it back.

[hallow thistle]

I still don't understand what you're tryna say. Do you mean like TTS, image prompt or an LLM model?

[long oyster]

i scan with this and it remove the app

[hallow thistle]

[stiff knoll]

i send you a message because i am new to ai and dont kniow many things to fully explain in text

[hallow thistle]

If neither of what I said is what you're looking for, then I don't know what you're looking for, sorry. That's my best to understand.

[stiff knoll]

so basically i tell the promt to write a text about 1000 words

and it writes 400

and whebn i try again it continues

[long oyster]

did i do something wrong?

[simple ore]

LLM models, especialy you can get for free, have a cap on the number of output tokens they generate

[stiff knoll]

okay thenx for the help

[hallow thistle]

Did you allow it back yet?

[dapper hull]

is the deiteris fork website down

[hallow thistle]

I think something is wrong with your internet or firewall.

[simple ore]

it may be blocked in your country

[hallow thistle]

If Rentry is blocked in your country, use VPN.

[simple ore]

or if you just want to download the voice changer, it is on huggingface

[wheat copper]

kept trying to load a new model and it does this

[dapper hull]

okay thanks!! it worked with a vpn

[low shard]

or just use our ai hub docs https://docs.aihub.gg/w-okada/local/w-okada/

it has the same guide

gonna change the -realtime wokada deiteris fork guide with the docs one

-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

[low shard]

yup it works

[hot lagoon]

Was a good amout of data to use batch 8?

[restive plinth]

hii im having troubles with okada, it used to work pretty good andd when i left for couple hours the voice changed to a sharper worse voice?

[low shard]

Elaborate

What's your PC GPU? What guide link?

[restive plinth]

nvidia geforce gtx 1050 and i used https://www.youtube.com/watch?v=LX5en3pZJwM&t=307s&pp=ygUVamFyb2RzIGpvdXJuZXkgd29rYWRh0gcJCYQJAYcqIYzv

[olive plover]

hello

[low shard]

Don't use YouTube tutorials

[olive plover]

FORK download link
?

[low shard]

It's using ah old version of original wokada

[restive plinth]

oh okkk

[low shard]

Plus Beatrice is lower quality than RVC, just lighter

[olive plover]

Is FORK the best so far and is there a new update for it?

[restive plinth]

the problem is i sounded smooth and fine the other day but when i restarted it same settings but sounds different lol

[low shard]

@restive plinth never use video tutorials for ai, a year old info is like prehistorical in ai field

[restive plinth]

okay i see do you have a good guide for setup?

[low shard]

Uninstall everything 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

[olive plover]

?????????

[low shard]

1st link

[elder coral]

-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

[low shard]

Wokada deiteris fork

Has way better quality and performance

[restive plinth]

okk!! ill try

[low shard]

Have patience, I can't talk with multiple people at the same time

[olive plover]

I hope for a response

[olive plover]

okay sorry🩷 <3

[low shard]

The Wokada deiteris fork is still the best, last update was 7th December 2024

[hot lagoon]

is this valid cuz there is new one's
so I would like to know what new one's I shoud use

[low shard]

I'm guessing you're talking about wooada deiteris fork when you said fork, since fork just means modified version in IT

[low shard]

Wdym

[hot lagoon]

Never mind

[olive plover]

Is there a new update for it? Or will it remain on this version?

I want the download link

the site

[low shard]

May be better to ask @crude flame since he takes care of the docs

[low shard]

As of right now there's no update from the one I said, we don't know when deiteris will make a newer version

[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

[olive plover]

@low shardthx🩷

[low shard]

Yw

[olive plover]

Is this site new?

[wheat copper]

Could anybody help

[olive plover]

How about this?

@low shard

[low shard]

No, it's our official docs since a while

[low shard]

It's the same guide as the one in the docs

We just moved it to the docs along with our other guides

But the only difference between those 2 are just the link lol

[olive plover]

oh okay

[olive plover]

@low shard

what i do?

[restive plinth]

oh yeah its a website

strange

[low shard]

Allow it

[heady tendon]

this site runs on your computer, also you need it to convert your voice

[low shard]

Else the program wouldn't be able to use ur microphone to change ur voice

[low shard]

Please read https://docs.aihub.gg/w-okada/local/w-okada/#why-does-it-run-in-a-browser-and-not-its-own-window

[tulip cloak]

That's not a website

[olive plover]

i do ( Allow while visitine the site ) okay?@heady tendon@low shard

[restive plinth]

oooh i see

[low shard]

Yes

[twilit hedge]

Hi can I get help

[low shard]

Yes, elaborate

[twilit hedge]

I'm trying to use okada in google colabs and it's giving me this error

ModuleNotFoundError: No module named 'pyngrok'

[low shard]

Elaborate:

• what's ur PC GPU
• what you want to do
• which google colab link are you using

[twilit hedge]

Intel
Use a voice changer
ts https://colab.research.google.com/github/w-okada/voice-changer/blob/master/Realtime_Voice_Changer_on_Colab.ipynb#scrollTo=lLWQuUd7WW9U

[low shard]

Which Intel? You can check in task manager

Also, that link is broken

There's Hina mod Wokada deiteris fork but that's broken too

There's wooada deiteris fork (a modified version which is better) but it gets detected that it uses a Web UI I free tier which isn't allowed so it will work only if you pay for colab

Don't follow video tutorials for ai btw

[twilit hedge]

Alright thx

[low shard]

What's your PC GPU?

I can help with other ways

If it's good enough you can run it on your pc

[twilit hedge]

It's Intel UHD Graphics 630

[low shard]

google colab is a cloud computing service only for people with a bad PC to use good remote pc

[twilit hedge]

I got nothin better

[low shard]

Cooked, your only other options are:

• buy a better PC
• use Kaggle 30 hours weekly but needs a phone number verification by google
• pay for colab

https://www.kaggle.com/code/suneku/voice-changer-public this is the Wokada deiteris fork kaggle

[twilit hedge]

I basically live in the streets 🥀

[twilit hedge]

Thx

[low shard]

Well, GPUs are expensive asf, no service is gonna give u them for free unlimited

[tulip cloak]

Damn man, u gotta updrage soon

[low shard]

It's already rare there are free ones

[tulip cloak]

I saw weights on whop. Com

[low shard]

Should have already upgraded time ago

[low shard]

Yw Let me know how it goes

[restive plinth]

is there a fix to the voice completely breaking when i start a game?

[low shard]

What game? Be sure you set the settings to the MINIMUM then show a screenshit of wokada while you start the game too

[tulip cloak]

yes show a screenshit

[restive plinth]

overwatch, all my settings are lowest

ill send a screenshit

[low shard]

Dang GTX 1050 ti

Also the settings are wrong

[restive plinth]

yeah it be like that sometimes broke hours

[restive plinth]

oh okk

[low shard]

Click stop, set f0 to rvmpe, set extra to 0.3, uninstall vb audio cable and get vac lite from the 3rd step of the guide then set the chunk to a value a bit higher than the chunk

@restive plinth try this and lmk

But I doubt it will run much fine tbh, also close useless programs in background

You can't expect too much from such old gou

[tulip cloak]

@low shard

What gpu you got

[low shard]

RTX 4060 ti 16gb why

Wbu

[restive plinth]

are these settings correct now?

[tulip cloak]

Oh, it's nothing. I was just curious. I got a 3080 if I'm not wrong

[knotty moth]

the whole pc market is expensive

[tulip cloak]

I mean

You still gotta keep up

[knotty moth]

especially the gpus

[tulip cloak]

You are maki danye right

My guy just disappeared

[knotty moth]

https://tenor.com/view/huh-cat-gif-26460616

[low shard]

Those settings are better, if you want less delay try https://docs.aihub.gg/w-okada/local/w-okada/#reduce-more-delay

[viscid moss]

Nice, ty!

[wheat copper]

Keep on getting this error when installing applio for amd

I followed every instruction

[junior iron]

been using a version of okada for legit a month now, suddenly I cant select an input? No matter what I try select it just wont change from none, but its perfectly fine when I use server

[simple ore]

well, you did not

[wheat copper]

How

What did I miss

Please help

[simple ore]

the path

[keen mountain]

I’ve never used applio I wanna install it but idk how to make models on applio can anyone help

[low shard]

what's ur pc gpu?

[wheat copper]

Im pretty sure I did that

[low shard]

elaborate:

• what's ur pc gpu
• what guide link did u use
• what browser are u using
• a screenshot

!give-media-perms 1h @junior iron

damn your bio reminded me of the game "to the moon" but it's actually a quote from edge runners

[simple ore]

show me how you added that

[wheat copper]

Wait nvm I did it wrong I didnt do new on path

I js added it to something else

I did thst now it works

[brittle wing]

yall what voicechanger type is this ? #1256387296677072988 message

[wheat copper]

Wdym

[brittle wing]

i downloaded it and it only has .pth file

[wheat copper]

Well yeah

I guess they just don't have an index?

You can still use it

[hallow thistle]

That is RVC voice model.

[brittle wing]

what does the index do thoe ?

[hallow thistle]

An index file stores accent of a voice model. Without it, the voice model will work fine but without accent applied.

[brittle wing]

aaa ok

[brittle wing]

and do you have maybe any recommendation for the settings i got 2070 super 8gb

[hallow thistle]

W-Okada? Are you following this guide? https://rentry.co/ForkVoiceChangerGuide#download-for-nvidia-gpu-on-windows

[brittle wing]

yes, and no

i used to use okada before so i know how it works but i can't remember what settings i used

[hallow thistle]

If you have followed any tutorial video on YouTube before this one, you'd better stay away from that. These videos only telling you to install the original version of W-Okada, which is outdated.

[brittle wing]

i just found the okada on github

and downloaded it

https://github.com/w-okada/voice-changer/blob/master/docs_i18n/README_en.md

this

[hallow thistle]

That's the GitHub repo for the original version of W-Okada.

This is the better one. https://github.com/deiteris/voice-changer

[brittle wing]

alr

thank you

iam downloading it right now

with one do i choose?

[hallow thistle]

Download both of them. When extracting, the program will combine both of these zips into one.

[echo meadow]

which ai collab thingy are there?

like collab

or something

[low shard]

What

[low shard]

Google colab is a cloud service provider for running code on remote good PCs, made for people with a bad pc

Anyone can write code for it

Specify your PC GPU and what you want to do

[echo meadow]

i got good pc, what do you recommend

i want to do ai cover

[low shard]

What's your PC GPU

[echo meadow]

4070 super 12gb

[low shard]

Great, get Applio (rvc fork, modified version) from https://docs.aihub.gg

[echo meadow]

hm wheres the applio

found

i dont need to train models, only want to make ai cover of it

[junior iron]

sorry got caught up, with something

[echo meadow]

just saying

[low shard]

Yes, and applio can do both

[junior iron]

i have an i7 9700k, 3060, not entirely sure of guide link since its been a minute

[echo meadow]

this is the best one so far?

the applio

[low shard]

Nice, share a screenshot of the program and also say your browser name

[low shard]

Yes, it has performance improvements

[junior iron]

legit its just input that wont change no matter what i select, and im using opera

[low shard]

Share a screenshot of the entire program

I feel like you're using an old original wokada from some yt tut

[low shard]

Because they advertise vb audio cable, which is known for randomly stop working

[low shard]

Yup that's the issue

Uninstall that old original wokada + the vb audio cable

never follow yt tuts for ai, even a year old one could contain prehistorical info

AI progresses at Sonic speed, you always have to get up to date

-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

Way better performance and less bugs

[junior iron]

ah alright so just wipe everything clean and use that guide

[simple ore]

"im using opera" ... yeah, no good

[junior iron]

is opera really that awful?

[low shard]

Yup

[echo meadow]

Before I use it locally on my pc the applio, do I need C drive for it? Cuz mines is almost full and has 117gb while my E drive has 2TB

[low shard]

Opera is also known for giving issues on Wokada, and it's known on the internet for security issues along with eating resources bc of their fancy effects

[simple ore]

15 years ago it was decent, nowadays after they sold old to god knows who

[low shard]

Should work on other drives too

[knotty moth]

not only that but ppl also reported some voice changer issues on it

[low shard]

They got a weird amount of browsers lmfao

Opera, Opera Air, OperaGX, Opera Mini

I don't even know how many other they have

[knotty moth]

opera has been owned by another entity with hidden spywares

fyi the former opera founder also made Vivaldi

[junior iron]

alright lads, well il give this new guide a look and hopefully all comes good

[brittle wing]

can i use VB cable on this one ?

[low shard]

Vb audio cable gives random issues on windows

Don't use it

[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]

Follow the 1st link, Wokada deiteris fork

Get vac lite and latest Wokada Deiteris fork

[low shard]

Yw and lmk

[brittle wing]

btw what do you recommend i was thinking of building a second pc to only run the voice changer and iam going to use to make calls on g voice to troll people w girl voice and i don't want it to like lag while talking you know

[low shard]

Yeah of course you can use a 2nd GPU only for the program

That would be good

[brittle wing]

what kind of gpu should i get

[low shard]

You can get even an RTX 3060 for main and then use the 2070 super for the program for example, but it really depends all on your budget and how less delay you want

[brittle wing]

ok

thanks

[low shard]

Really also depends on what games you play btw

I wouldn't make a PC only considering the voice changer, consider your other use cases

[brittle wing]

ye iam running dma

so

what settings should i use if i got 2070 thoe ?

[tulip cloak]

Nick really works overtime

[low shard]

80 Ms + extra 2.7

If youre gonna game, just put the chunk higher than the perf value

[brittle wing]

like this ?

[low shard]

yes

[brittle wing]

do i got to change anything in advanced settings?

it so diff from the original okada

[low shard]

only force fp32 mode for better quality and stability, but it adds more delay

[brittle wing]

ok

ima going to try these settings right now

[brittle wing]

if i remember correctly if its a girl model i have to lower the pitch right?

[low shard]

lower pitch = deeper voice
higher pitch = higher voice

[knotty moth]

depend on how high is your voice

[brittle wing]

ye i realised after sending the message

i got a brain freeze

[junior iron]

oh yeah, it works now!, thanks a ton lads

[brittle wing]

thank you nick this voice changer is so good better than okada

ngl

almost 0Delay

[low shard]

may u want me to check your settings?

[brittle wing]

but on discord it sounds pretty weird but on the voice changer monitoring sounds better

[low shard]

maybe try https://docs.aihub.gg/w-okada/local/w-okada/#discord-crackle-fix

[brittle wing]

tried it

its not like cracking

but sounds like iam in a tunnel

or something like that

[low shard]

elaborate more

[echo meadow]

i remember there was something that you could get all inclusive things like, main vocals, background vocals etc

all in one

[brittle wing]

give me some time

and i will send u the voice message

so u can see

[low shard]

https://docs.aihub.gg/rvc/resources/dataset-isolation/

[junior iron]

yeah sure

[echo meadow]

hm cant find that 1 simple model called Gabox's voc_fv4

maybe its outdated?

#The best models for UVR are:
Extraction Process Method Model
Vocals MDX-Net Gabox's voc_fv4
Instrumental MDX-Net INST Gabox V7
De-Reverb VR Anvuew mel dereverb v2
Extract Backing Vocals VR Mel roformer karaoke
De-Noise VR Mel denoiser

[low shard]

are you using the beta?

[low shard]

!give-media-perms 1h @junior iron

send a screenshot

[echo meadow]

idk

but last time i used one of them, it was like collab and when the stuff are done like model download and songs etc, it got like automatically did for me for background vocals, main vocals etc

[low shard]

you could optionally turn force fp32 mode for better stability and quality but worse delay

also, i meant normal settings too

[low shard]

what are u using

[echo meadow]

applio rn, but last time i used collab

for like 2 years ago

[low shard]

applio isn't related at all with cleaning the vocals though, you'd need another program

[echo meadow]

ah i meant i used ultimate vocal remover 5

[low shard]

i think it was AICoverGen, another RVC fork, which is pretty outdated, the only updated rvc fork that does that automatically is: https://github.com/Eddycrack864/RVC-AI-Cover-Maker-UI

[echo meadow]

that does at the end like automatically backgorun vocals, mainvocals, insturmental etc

right

[low shard]

yes, it's only for ai covers though

there isn't a guide except the one said in the github repo

[weak island]

are there tutorials in this discord. I'm looking to try to make an AI cover training a model with my voice, i wanted to play around with it. Can anyone point me in the right direction?

[keen mountain]

Idk it’s a laptop

[low shard]

a laptop is a type of pc

[keen mountain]

Ik

[low shard]

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

[low shard]

could you firstly tell me your pc gpu? just so i can tell you what guide to use

[keen mountain]

It says “Inter(R) UHD Graphics”

[low shard]

is there any other gpu?

[keen mountain]

I don’t think so

[low shard]

damn you're cooked

your laptop is too weak to do it locally

but you can use cloud

Train (make) RVC Models on cloud:

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

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio (UI)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • 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

[keen mountain]

Alr

[tulip cloak]

Damn man

[low shard]

mainline is broken rn btw

[low shard]

I gave you the cloud methods

[tulip cloak]

When is hina fixing it

[keen mountain]

Wha do I pick

[tulip cloak]

Been waiting for ages

[low shard]

who knows

they are busy

could be some other months

[tulip cloak]

Yeah I respect it

[low shard]

Please read carefully, I explained the differences

[keen mountain]

Ok

[keen mountain]

I think collabs would be the best

[low shard]

alright, applio colab might be good for you then, just reminder that the gpu time isn't much ideal for model training unlike applio kaggle

[keen mountain]

Will I still be able to train models on it

?*

[low shard]

yes, but be aware that the gpu time is random and lower, meaning that there could be issues about getting disconnected mid training

tho it's easier to setup

[weak island]

6800xt

[low shard]

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

[keen mountain]

For downloading my models it says to have a hugging face or google drive link but I haven’t uploaded anything on huggingface

[weak island]

working through this now, thank you brother man

[keen mountain]

Nvm I think I figured smth out

[weak island]

i cant link images in the channel but i got an error installing the amd 6.12 HIP thing

Error 186 – AMD Software Installer Cannot Continue Due to Missing or Corrupted Manifest Files

searching around now for solutions

[swift thunder]

Help, I get this in Applio Colab

[keen mountain]

Ima js give up it’s too hard

Fuck applio

https://tenor.com/view/esports-game-game-time-game-day-game-on-gif-20437016

[low shard]

what?

elaborate the issue

send a screenshot

[weak island]

On the step "Using command line from the Applio folder run" which command line? in the cmd from the .bat?

or in the browser window that opens up

[low shard]

just go to the applio folder, write cmd in the windows explorer at the top bar, then run

@simple ore btw this might be a bit more specified

[weak island]

just go to the applio folder, write cmd

[viral ivy]

Should I be using
https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md

or w-okada
2070 super 8gb vram

Is the GitHub just a lighter weight software?

Trying to do real time

[tulip cloak]

Wait how did ur 3070 become [e-mail]

[viral ivy]

I have two computers one with a 3080ti and another with a 2070 super

[tulip cloak]

Wait you are looking to do real time

Then go for wokada

Not for rvc

Nicks will come and help you out soon

[viral ivy]

Ok sounds good I couldn’t figure out what the GitHub’s software even was

I should be able to figure out wokada

[tulip cloak]

I mean it's pretty easy

[viral ivy]

The GitHub was just confusing on top of the translations, over my pay grade and level of patience

[tulip cloak]

So you better

[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

Wokada = uses RVC for realtime inference

what do u need

[viral ivy]

Oh easy pz I got it then

Thanks nick

Wasn’t sure if RVC was just a complicated way to do realtime lighter weight

Didn’t fully understand it

[low shard]

i mean, rvc has a go-realtime.bat, but it's not as good as wokada deiteris fork in performance terms, also it's more messy and buggy

[viral ivy]

Cool got my answer then, ty !

[analog obsidian]

the new one is better but is extremely buggy and hard to install

[low shard]

what you're talking about isn't the new one, its the fumiama mainline fork, which works just for you lol

[analog obsidian]

yes lol

[viral ivy]

Yea I skimmed the readme through the GitHub and was like this sounds good

The fork

But didn’t want to go through the effort blindly

Normie way should cut it for me

[low shard]

please always mention it as fumiama, it's different than mainline

don't want users to get confused

[viral ivy]

Bought to make a killing on TikTok /s

I’ll remember you guys

[crude flame]

💀

[analog obsidian]

[low shard]

what?

[viral ivy]

Kidding

[crude flame]

rvc aint that good

[analog obsidian]

rvc is good until u laugh and show human emotion

[crude flame]

😔

[viral ivy]

Surprised how far it’s come in a years time since I last checked

Ran a handful of your models through a recording of me on a phone call @crude flame

Really well done

Either way I’ll stop clogging support

Ty again

[tulip cloak]

You remind me of someone

[vivid forge]

Is there a software that can seperate duel singing duet vocals? The isolate Main vocals option on Weights.gg is shit and keeps the two dueled vocals together when when Isolate it and get backup vocals.

[glad escarp]

how to upload my model in the channel voice models

[upbeat carbon]

hi everyone!

hope your having an amazing day!

and stay safe!

[opal cobalt]

hello everyone did someone know how to fix this error when i try to train on applio pls ?

[simple ore]

well, it says what it says

you either forgot to slice it, or sliced your self too large or too small

so there's nothing collected into the training set

[wheat copper]

What's considered too large

[simple ore]

10s

really >3s

[wheat copper]

Ah ok

[wheat copper]

What about for a whole dataset

What's considered too much for that

[simple ore]

smaller than 10 min may need some tweaks to the training parameters

more than an hour would provide no useful benefits

variety >> quantity

[wheat copper]

Thanks for the answer, now I gotta get rid of like 30min of data lol

[limber vessel]

When i downloaded voice.ai im stuck behind a subscription pay screen and i dont know how to get rid of it

[tame mica]

we dont give support for voice.ai

[limber vessel]

okay

[tame mica]

💀

mf left quick asf

df

[opal cobalt]

thx you i don't have this error anymore but i have another did you know a way to fix it pls ?

[brittle wing]

does anyone know why it sounds like this?

[wheat plover]

How do i download custom UVR models? I was looking at the Bleedless vs Fullness spreadsheet by Bas Curtiz and was wondering how I could use some of those models.

[toxic nacelle]

hello hello,, ummmm may anyone help me how to install and setup the voice changer, i've been trying for months and noone helped me\

[viscid moss]

https://discord.com/channels/1159260121998827560/1343297771104501871

[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

[viscid moss]

@toxic nacelle first link

[toxic nacelle]

got it worked, thx to @crude bolt

[wheat plover]

I'll check it out, thanks

[worthy fox]

hi, how can i change the echo after the voice is changed? https://www.weights.com/shared/cma5wk9ba0gylp0159e009kh1?inviteCode=6d973c

[simple ore]

did you forget to run preprocess/extract features?

[wheat copper]

im trying to train models on my amd system but everytime i made a model it doesnt work

why is that?

it gets an error

[toxic nacelle]

ok, so i got the ai working on the website, but not in discord and vrchat , pls consider @ me

[whole crypt]

if anyone ever has problems using the w-okada mergelab feature again, i think i figured it out: i had copied my old model_dir from one version to another. for some reason that seemed to break everything, but reinstalling w-okada branch and uploading every model through the GUI itself worked! i think maybe when you do that it creates data files for them that aren't visibly stored in the folders themselves or something? no clue why, but it works.

[hallow thistle]

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

[knotty moth]

bro wanted to pay for voice.ai

[hallow thistle]

I'm not sure what you mean by "how can I change the echo after the voice is changed". I think you mean this feature where before you start creating your AI cover on Weights.

[hallow thistle]

One sentence just made him cried crazy.

[whole crypt]

the most recent version of the fork (b2332), i have a radeon 7800XT.

[hallow thistle]

Did you follow this guide? https://rentry.co/ForkVoiceChangerGuide#download-for-amd-intel-and-cpu-on-windows

[whole crypt]

oh, my problem is resolved! i don't need help, but thank you. yes, i followed that guide, but i copied my model_dir folder over from the non-fork version. that is what i believe caused my issues yesterday!

(because using the same files but not copying the folders made the subsequent merges work, for some reason.)

[hallow thistle]

Case solved.

[whole crypt]

yup! the folks that helped me out said it was a new bug though, i just wanted to leave a message in case someone else stumbles into it!

[knotty moth]

the models in model_dir folder of any voice changer versions are actually converted ones

so you should first upload models to your current version to avoid conflicts

[hallow thistle]

Although, I'd say renaming voice model name on fork W-Okada UI is bugged. The only way to rename the voice model is to go through model folders and use Notepad to edit it.

[terse nymph]

No longer have access, where do I find the guide for it

[tawny raptor]

fellas

What is the most intelligent, most up to date chat model out there? I am not familiar with chat models

I need a chat model to organize business and I don't want to use chatgpt for privacy concerns, but I want a model of nearly the same intelligence or better

I am familiar with the signficance of the original pretrain for voice cloning in this github page https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI

What is an equivalently signficant pretrain and ui for chat models?

[hallow thistle]

-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

[terse nymph]

Tanks

[hallow thistle]

[fallow quarry]

Hey so I'm trying to use the mainline RVC colab but when I click the RVC link I get this error; what should I do?

I can't paste the screenshot somehow

"ERR_NGROK_8012

Traffic was successfully tunneled to the ngrok agent, but the agent failed to establish a connection to the upstream web service at http://localhost:9303. The error encountered was:

dial tcp 127.0.0.1:9303: connect: connection refused"

[ornate oracle]

hey I'm trying to make a shucks retold cover with my oc's demon form, jake, elliot and milly and I'm trying to make ai covers of the voicelines but it just says errored

here's the error

Errored: Failed to load audio file /dev/shm/in_memory/cma6fe5370otskv15qflnvztm/cm38huhq31oal5ed9csqujs8h_b171b8fb-53e7-4fd0-92f6-b7df705252ba.mp3: ffmpeg version N-71064-gd5e603ddc0-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers built with gcc 8 (Debian 8.3.0-6) configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg libavutil 59. 27.100 / 59. 27.100 libavcodec 61. 9.100 / 61. 9.100 libavformat 61. 4.100 / 61. 4.100 libavdevice 61. 2.100 / 61. 2.100 libavfilter 10. 2.102 / 10. 2.102 libswscale 8. 2.100 / 8. 2.100 libswresample 5. 2.100 / 5. 2.100 libpostproc 58. 2.100 / 58. 2.100 [mp3 @ 0x11763040] Format mp3 detected only with low score of 1, misdetection possible! [mp3 @ 0x11763040] Could not find codec parameters for stream 0 (Audio: mp3 (mp3float), 0 channels, fltp): unspecified frame size Consider increasing the value for the ...erminating thread with return code -22 (Invalid argument) [aost#0:0/pcm_f32le @ 0x11768140] Could not open encoder before EOF [aost#0:0/pcm_f32le @ 0x11768140] Task finished with error code: -22 (Invalid argument)

[aost#0:0/pcm_f32le @ 0x11768140] Terminating thread with return code -22 (Invalid argument) [out#0/f32le @ 0x11767700] Nothing was written into output file, because at least one of its streams received no packets. size= 0KiB time=N/A bitrate=N/A speed=N/A Conversion failed!

[tulip cloak]

TUNG TUNG TUNG SAHUR

[ornate oracle]

tf? can you help plz

[tulip cloak]

Oh my bad, I wasnt replying to you

[proper oasis]

Whats the ai so i can transform like from PHOTO A to PHOTO B
Example: https://www.instagram.com/reel/DJETRqtK1v7/?igsh=MWszdTI2OTllMnEzdQ==

[low shard]

What's your PC GPU? What tutorial link are u using?

[low shard]

Because it's broken since months, always check #📰│dev-updates

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

[low shard]

Oops old message, unpinned it

[viral ivy]

hey nick could you help me out here

i have everything setup right i believe

ran through a text guide

but when I try to run through server i'm not getting any output on my virtual cable

seems im not getting any connection?

when i try to do client i'm cooking my CPU and my GPU isn't being used at all

causing wild latency

should i be using a fork or something to use more gpu/less cpu?

2070super and a i7 9700k

doh..

thatll do it huh

Should force fp32 explode my pc

Windows explorer crashes and the web extension thingy is just stuck on waiting on operation

[latent moat]

I'm not sure how to avoid blue screens with the voice changer

[viral ivy]

You’re probably just cooking your hardware if it’s lower end tbh

Surprised how demanding this is lol

[latent moat]

Running ryzen 7900x, 7900xt, but I commonly will experience blue screens strictly when having the application open. This is during heavy use, e.g. VR.
windows 11

[outer wasp]

Hi, Why voice changer client keep stop working?

[tulip cloak]

This is real firepower

[ornate oracle]

an intel R hd graphics 520

[latent moat]

your battle with a integrated GPU on a voice changer will be legendary, NDALT.

give up.

that's 8 years old

[ornate oracle]

OH COME ON!

[latent moat]

you're trying to go on the highway in a 1930s car

[ornate oracle]

lol

[knotty moth]

but can it run crysis?

[low shard]

That's Crazy too weak, but you can use cloud (remote good PC)

What do u want to do?

[low shard]

Elaborate pls

[low shard]

Close useless programs and lower the game graphics to minimum, be sure to be using Wokada deiteris fork b2332

[latent moat]

Yeah, been running that version for a bit. Idk about lowering game graphics as it's VR, i can lower render res but I ain't close to 100% GPU usage as told by the driver..

I'll reply to you if that fixes it though.

[outer wasp]

So i use Okada VC, when i use first 5min its okay. And then, it just silent

[knotty moth]

your windows seems to be possibly fucked up, try updating or clean install it

[low shard]

What's your PC GPU? What tutorial link did you use?

Send also a screenshot of your wokada settings

[latent moat]

i mean i guess

[outer wasp]

NVIDA 4060

[low shard]

!give-media-perms 1h @outer wasp

Be sure you didn't use video tutorials since those are old

[knotty moth]

or try updating the gpu driver

[low shard]

Nice

[outer wasp]

Okay wait min!

[latent moat]

nah, they fucked up VR use on 25.4.1 25.3.1 25.3.2 on amd drivers, still running the february one 25.2.1 because new shit just crashes after 3-5h of VR use, memory leak
it's in the "known issues" too

[outer wasp]

And what VC everyone use the most? Cuz i think Okada is not stable for me

[knotty moth]

the issue is only on VR right?

[latent moat]

yes, only along with VR use do bluescreens/crashes/force restarts happen

because i usually do VR use and unity editing, building at the same time, so i'm used to having PC under a lot of stress

[outer wasp]

i cant login oh no

[knotty moth]

you can check out on some AMD community forum

there could hopefully be some workarounds there

[latent moat]

there's like 9 results on google, this is way too specific

Especially because already a 2 year old forum post doesn't help as it's too outdated, and people that use real time voice changers are so rare on top of it.

[low shard]

Lemme guess, you used a video tutorial?

[outer wasp]

Yes…

[low shard]

Because that seems to be a very old version of original wokada

[outer wasp]

Ohh

[low shard]

a year old video is like prehistorical in ai field btw

uninstall all u got from 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

[low shard]

follow 1st link, wokada deiteris fork

[ornate oracle]

I have an Idea.

[outer wasp]

OMG thank u sm!!

[knotty moth]

well we'd wish AMD good luck on fixing the issue

[latent moat]

Right.. i'll see if high gpu load is the issue and if not i'll see if an older driver fixes it.

[low shard]

?

[latent moat]

his computer died from the load, hold on, he cooked

[fallow quarry]

I don’t have a GPU (and only have a laptop) so local usage isn’t an option for me :,)

[low shard]

btw laptops are pcs too, i’m guessing you already checked its gpu in task manager?

[low shard]

lmaoo

[fallow quarry]

Yeah
But it doesn’t seem to be usable for general tasks and stuff

[low shard]

i'm guessing it's integrated graphics

Train (make) RVC Models on cloud:

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

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio (UI)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • 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

ofcourse, mainline is broken right now

[fallow quarry]

Right so I just use one of the others?

[low shard]

Yes

[fallow quarry]

Will do
Thanks!

[sly furnace]

@crude flame so what should i do to make a perfect song with my own voice ?

[polar dove]

how do i train models on my pc

[opal cobalt]

ok my bad now it working it probably that sorry for asking for this simple probleme

[low shard]

What's your PC GPU

[polar dove]

NVIDIA GTX 1660 super

6gb

[low shard]

Simply, you don't train it on your PC

You could even do it locally but it would be slower and limited by VRAM compared to cloud

Train (make) RVC Models on cloud:

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

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio (UI)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • 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

Mainline is broken rn

[polar dove]

why did you delete the first thing

[low shard]

Wrong message, it was related to do it on phones