✨│ai-help

[echo turtle]

this everytime i click the convert button, https://www.weights.com/models/clm737ynf1tgqcctc4xnna57g = model

[low shard]

put extra to 2.7
put sr to 48000
did you follow the windows wasapi guide

[nocturne stone]

no i did not

[low shard]

you need it

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

check the 3rd step of the guide

[prisma dove]

ahh i have one

[low shard]

https://rentry.co/LessDelayWasapi

[low shard]

if you're talking about vb audio cable

that's worse, it gives random issues on windows

it's better to uninstall it

[prisma dove]

ok

[nocturne stone]

thank you

[prisma dove]

that correct one?

[low shard]

yw and lmk

[low shard]

the one you showed is vac lite, which is the correct one

you got it already?

[prisma dove]

yeah

[low shard]

alr, then set the output in wokada to line 1

and set the input in the other program (for example discord vc) as line 1

[prisma dove]

already done

[low shard]

also be sure you are using chrome or firefox so you don't risk running into issues

[prisma dove]

ok, i use chrome

[low shard]

now, upload any models, click start, check the perf value at the top left, and set the chunk higher than that value

like if it's 200, you set chunk to 260

[nocturne stone]

thank you so much, it's good now

[low shard]

you're welcome, any other issues?

[prisma dove]

when i uses headphone to listen myself everything clear. But in vc in discord with line 1 it has issues with ping etc

[nocturne stone]

question whats a good chuck to put it at

[low shard]

it's better you close apps in background, start, check the perf value then put the chunk to an higher value than that

like you get 200 perf value, you set chunk to 260

this is the way to find your own best chunk value

[low shard]

go into discord voice settings and turn off echo cancellation and noise suppression

[echo turtle]

hey do you have insight on what the problem seems to be?

[nocturne stone]

ok thank you

[echo turtle]

this could be a better picture

[low shard]

weird, could you try another model to see if it's a model issue

[echo turtle]

yeah

also are models from voice-models.com valid?

or is it outdated

[low shard]

as long as they are rvc v2 they should work

[echo turtle]

ok

[prisma dove]

Ill describe it when headphone sound is amuzing but in discord it very terrible

[low shard]

yeah did you try as i said

[prisma dove]

i can try to record if wanted Just the sound pretty crazy. I off everything as you say

[low shard]

alr post it here

[prisma dove]

maybe cause of trial version on vac?

[low shard]

did you download trial or lite?

[prisma dove]

trial

[low shard]

install lite

[analog obsidian]

the model was trained in 2023 using harvest, there's a high chance the model is a v1 one

[low shard]

thanks for pointing out

[echo turtle]

ahh

let me try a different

one

what is a ckpt file?

[low shard]

that's a checkpoint file

where tf did u even find that

[echo turtle]

some model in the voice models forum here

[low shard]

that's a gpt so vits model, not rvc v2

[echo turtle]

oh

[low shard]

not all voice models are for rvc v2, those for gpt so vits have the gpt so vits tag or in the title

[echo turtle]

sorry im kinda new to this lol

[low shard]

basically

[echo turtle]

ahh okay

[low shard]

there's different types of ai voice models

gpt so vits is tts (text to speech)

rvc v2 is the best sts (speech to speech)

so vits svc is an old sts

[echo turtle]

ah okay

is the index required

this model didnt come with one

its for rvc v2

[prisma dove]

Now everything fine. Thank you very mutch

[low shard]

no, it's optional

in rvc context:

• pth is the voice
• index is the accent

[echo turtle]

ah okay

[low shard]

You're welcome, any other issues?

[prisma dove]

no ty

[low shard]

have a nice day

[river cairn]

Just wanted to bump this. I'd appreciate any input you guys have!

[low shard]

Wokada deiteris fork is the best as of right now

What's your PC GPU?

Also be sure to uninstall the version you got and vb cable if you had it

[river cairn]

NVIDIA RTX 3050 Ti Laptop

I've got Muzychenko's VAC, I remembered not to get VB Cable from my original installation haha

[low shard]

You won't have issues except for high end games like marvel rivals basically

[low shard]

Great then

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[echo turtle]

it went through this time but still isnt allowing me to download it or listen to it

[low shard]

1st link, read up the guide

[low shard]

Maybe wait for a few seconds till it appears, is the audio file big?

[echo turtle]

about 7.3 mb

[low shard]

is the file output still not showing in the UI?

[echo turtle]

yeah

[simple ore]

manually download the model and unzip it into Applio/logs/homer

[low shard]

go to the applio folder then assets/audios/notlikeusvocals.mp3, you should be able to see it from there

[low shard]

the model might have been an rvc v1 one since it was trained on the era of when harvest was used

[simple ore]

well, v1 is usable

the error was about broken .zip (.pth)

[low shard]

wasn't v1 depracated from applio?

[simple ore]

can still infer v1

[echo turtle]

just the original vocals

this always happens to me

i always be creating this newly found error

lmao

[low shard]

oh is it removed just for training then?

[low shard]

weird, could you try to convert again? also follow what noobies said

[echo turtle]

tried converting again, nothing showed, tried rebooting applio, nothing showed, tried what noobies said, nothing showed

wtf ???

oh forgot to check console

[simple ore]

what model you're trying to use?

[echo turtle]

https://www.weights.com/models/clqhr3pe300773waybjufj7zh

[simple ore]

and you're using 'trained' index

not 'added'

[echo turtle]

it legit just says model.index

bad model?

[simple ore]

some idiot uploaded a wrong index

dont select it

[echo turtle]

okay

used the find models thing and this is the one it gave me

[simple ore]

that looks okay

anyway, the other model is fine

just use pth

[echo turtle]

finally worked

tysm

sorry for the trouble

lmao the start whisper

[simple ore]

you need to use voice separation model

[echo turtle]

how do i do that

[left wadi]

sorry for disappearing I went afk im back now just installed everything

[low shard]

alright, might want me to check your settings or if you encountered issues?

[low shard]

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

[left wadi]

I havent opened anything yet I just installed everything

[echo turtle]

👍

[low shard]

alright

[left wadi]

Whats vaclite?

you told me to install it I installed it

[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

[left wadi]

Oh

I see

To open wokada do I open this?

[low shard]

yes

[left wadi]

it opened as a browser

is this normal?

now that its open whats the next step?

[low shard]

yes, it uses a web user interface like the majority of ai programs

[low shard]

gpu: your rtx
f0: rmvpe without onnx
extra: 2.0
input: microphone
output: line 1

add a model

[left wadi]

idk why I cant find my microphone on my input

and I also cant seem to find line one on output

unless this means line one?

[river cairn]

Just to clarify, is there a particular method for uninstalling my old w-okada VCClient or do I just delete the entire MMVCServerSIO folder?

[low shard]

what browser are you using

[left wadi]

arc

[low shard]

just delete the folder

[low shard]

arc?

[left wadi]

yes arc

[low shard]

oh just googled it, seems like it doesn't get updates anymore

we suggest chrome or firefox to not run into issues

also be sure to give microphone permissions when asked

[left wadi]

alright lemme install chrome give me a second

okay done but I still dont know what you meant by line one

Holy fucking shit this feels so much better to use than the ancient one for some reason

[river cairn]

For installing the Deiteris fork, the instructions on the official Github (https://github.com/deiteris/voice-changer?tab=readme-ov-file#for-nvidia-users) and on the online guide (https://rentry.co/ForkVoiceChangerGuide#download-nvidia-on-windows) are not quite the same (at least superficially). Which one do you recommend?

[slender osprey]

Hey, my PC has 16 GB of RAM, Intel Core i5-8400 Processor, 2.80GHz CPU, 8 GB video card and 447 GB of storage, my internet is 100 MB, what can I install and configure with this?

[low shard]

For input you will see your microphone

For output you will see something called line 1

[low shard]

Is it all fixed?

[left wadi]

Cant find that tbh

[low shard]

Did you install vac lite as said in the guide?

[left wadi]

I did I just didnt use it because Idk how to or what to do

[low shard]

What's your PC GPU name? And what do you want to do exactly

[low shard]

So you didn't open the setup64.exe?

[left wadi]

no

[low shard]

U have to

[slender osprey]

I don't know, how do I see this?

[low shard]

Please read all the guide, read carefully step 3

[low shard]

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

And what do u want to do

[slender osprey]

GPU: AMD Radeon RX 580 2048SP

[pliant crest]

raceback (most recent call last):
File "/content/Hina_RVC/src/webui.py", line 8, in <module>
import gradio as gr
ModuleNotFoundError: No module named 'gradio'

Hina is not working

[simple ore]

likely uv install issue, add --prerelease if-necessary-or-explicit key to uv pip install lines

[pliant crest]

idk shit about coding

Idk where that goes

[simple ore]

find the requirements install cell, click

there's gonna be a few lines that start with !uv pip install

add --prerelease if-necessary-or-explicit after install, make sure there's a space

!uv pip install --prerelease if-necessary-or-explicit whatverelse_is_there

[river cairn]

Is there a recommended location that I should extract the Deiteris fork ZIP download to, or can it be placed anywhere in my local file system?

[simple ore]

C:\voice_changer\ will do

[pliant crest]

I still got an error

[simple ore]

or any other drive, not C:\user\blah-blah\onedrive\copy (2)\voice changer

[simple ore]

re-run the install cell

[pliant crest]

I did

still error

This did not happen before

[simple ore]

show the output log for the install, but you need to remove all "clear_output()"

[river cairn]

Is C:\Program Files\voice_changer good? Or does it need to be in the root?

[simple ore]

just the root

preferably no spaces in the folder name

[river cairn]

Awesome thanks, and could I put it on a removable drive? Or does it need to be in a permanent drive on the PC?

[simple ore]

does not matter

[simple ore]

anyway, that colab is outdated and not supported most likely

see #📰│dev-updates message

[pliant crest]

Hina separated vocals 4 u and added the intrumental

is there any similar model?

[simple ore]

what I gave you

or local separator

[unique rock]

Noobies, how can I download the custom pre-trains? It won't let me anymore.

[unique rock]

How do I use the new applio? Everything has changed. I can't download the pre-trains. When I load the model's backup, it doesn't appear even when I refresh in the train tab. The slice simple doesn't appear, and it stops when I truncate silences. 😭

[river cairn]

After running the Deiteris fork for the first time, it tried to open the WebUI but got this error (my browser is Firefox). Is this a browser issue, and is there an easy way to fix it on Firefox? I'd prefer to use that over Chrome.

[barren garden]

yo

hwo knows the best settings for ai girl

[knotty moth]

set sample rate to the audio spetrogram cutoff, choose batch size 4 or 8 for longer dataset, choose KLM pretrain for ur waifu voice

[barren garden]

idk what tf you just said

😭

can i dm you so i can send a screenshot?

[knotty moth]

cuz you replied to someone else's comment about model training

[knotty moth]

nuh uh

!give-media-perms 1h @barren garden

[barren garden]

where is it

[knotty moth]

og wokada for directml is dogshit, get better one here https://rentry.co/forkvoicechangerguide

[barren garden]

and that will help me get better girl voice?

[crude flame]

sigh

[slender osprey]

My PC has 16 GB of RAM, Intel Core i5-8400 Processor, GPU AMD Radeon RX 580 2048SP, 2.80GHz CPU, 8 GB video card and 447 GB of storage, my internet is 100 MB, what can I install and configure with this?

[simple ore]

to do what?

[devout goblet]

can someone send me the download for it in english

used to have it then wiped my pc and cant find it

[knotty moth]

sounds like ur a clown doing april fools joke

[slender osprey]

Which version can I install and what settings can I put in the application

[broken shell]

What is the best alternative for easygui (which has been dead) that has a similar gradio? Or... that is just similar to what easygui was in general? 👉 👈

[fervent frost]

2025-04-13 04:12:12.8595541 [E:onnxruntime:, sequential_executor.cc:516 onnxruntime::ExecuteKernel] Non-zero status code returned while running Mul node. Name:'/hubert/feature_extractor/conv_layers.0/layer_norm/Mul' Status Message: CUDA error cudaErrorNoKernelImageForDevice:no kernel image is available for execution on the device how do i fix this

[knotty moth]

seems like unsupported gpu

[fervent frost]

ty

[upbeat sparrow]

after i download the model, i need something else ? okada ?

also i want TTS not voice changer if possible,this one https://discord.com/channels/1159260121998827560/1175430844685484042/threads/1179052385532903494

[simple ore]

to do what? make images? realtime voice change? non-realtime voice change?

[simple ore]

what's your gpu?

it wont work with a very old device like <1000 series

[slender osprey]

Change voice in real time, sorry I wasn't specific

[simple ore]

https://rentry.co/forkvoicechangerguide

[inner haven]

thank you for the link, though i did follow the trouble shoot and did the task manager tip but it still continued, its like a poppy crackle

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

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

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

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

__**BE PRODUCTIVE.**__ 🤝

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

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

[hallow thistle]

"RVC" voice model can be used with an RVC the regular voice chager program, it doesn't have to be W-Okada the realtime voice changer at all.

Applio has a TTS feature built into itself.

[hallow thistle]

You have RTX 50 series GPU? Because some stable packages made for RTX 50 especially haven't out yet. Which RVC program are you running by the way?

[hallow thistle]

If you mean by audio quality, then sure. But if you mean by an e-girl voice model, hell no.

[heady laurel]

Helppp please recommend best computer specs i should buy so ill start saving

[hallow thistle]

But I don't have any computer specs recommendation in my mind. You'll have to look for one that's at least decent enough for gaming but also in budget for your needs.

If you want at least decent GPU for AIs, NVIDIA GeForce RTX 30 series GPU is at least enough. RTX 40 GPU is decent, stable and super fast.

[heady laurel]

Can you give me like specs to look for ? Im very new to buying computer, i always use laptops so 😭

[hallow thistle]

[heady laurel]

How about this one

Is this enough specs ? To afford, it might take me like 6 months to save up lmfaoo saving as a student is near impossible these days

[heady laurel]

Anyone? Is this good or nah aaaaaaaa

[hallow thistle]

Are you sure you have learn more about how powerful these specs are?

[heady laurel]

So is it enough? Since i want no lag and i dont want the training to be slow or have any problems with the audio?

[hallow thistle]

I'm not a professional of gaming/workstation PCs, so I can't judge it to be enough, but I'd say it's at least decent enough to run games and even train an AI model locally.

[heady laurel]

Hii nick please noticee can you guide me with specs ?? Im financially dying but ill save up to meet the best specs can you guide me so i dont end up dying in regret if i bought the wrong one?? I want to make sure 😭😭😭😭😭

[hallow thistle]

Again, I can just say pretty much it. It's up to what you need, not me or anyone lol.

Any CPU, GPU, RAM and SSD released after year 2020 are at least decent. Anything released before 2019 is considered to be old.

[knotty moth]

it is more than enough for rvc training but it depends on your budget. you can consider lower but still recommended specs: RTX 3060 12 gb + ryzen 5600X or 5700X3D, others are fine

[pastel oak]

If you use vb cable then thats the isssue. Download vac lite from the guidd

[long gazelle]

can anyone help me with huggingface?

ive had this error for weeks

[low shard]

that's good enough, you will have issues only if you use intensive games like marvel rivals

-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 guide, wokada deiteris fork

[low shard]

yeah that

now set line 1 as also the input in the other programs (like discord vc)

[low shard]

maybe it's better you give chrome or edge a try

sometimes browsers could have issues, welcome to javascript

[low shard]

for what? there's over thousands of AI programs
reply to:

• what do you want to do
• what is your pc gpu

[low shard]

try Applio

[low shard]

what's your pc gpu btw

[low shard]

that one is for RVC v2

There are different Text To Speech (TTS) AIs:

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

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

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

You can check TTS in our tts index

With RVC Models:

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

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

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

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

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

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

• You could try another tts from our tts index and use the output as an input in rvc

[low shard]

it all depends on your budget and what you need

[low shard]

seems like an huggingface internal issue

either try changing the license to openrail/mit or just wait

you could also report this in the official huggingface server where there are official staff

[long gazelle]

do you have a link by chance?

[low shard]

did you retry if the error still persists first? I just tried myself and huggingface works fine

maybe refresh the page, or try another browser

[long gazelle]

still nothing

I have had it for a really long time

[low shard]

https://discuss.huggingface.co/ or google "huggingface discord"

seems like an issue on their end

unfortunately, we can't do much about it if it's a site error

[fervent frost]

gtx 750 ti

[low shard]

that answers your own issue, that gpu is prehistorical basically

too old to even be supported by CUDA, which is needed for wokada

the bare minimum would be a gtx 900, but your gpu is weaker than that and even too old

your only options are either:

• try cpu mode (which is probably gonna have a lot of delay)
• use cloud (remote good pc, with limited time in free tier because gpus are expensive)
• buy a good pc

@fervent frost most cloud methods are having issues, since cloud is very unstable

the only working one right now is Wokada Deiteris Fork Kaggle which gives you 30 hours weekly of good gpu, but is hard and requires a phone number

let me know for further issues

[fervent frost]

Thanks

[low shard]

yw and lmk

[hallow thistle]

It's crazy you have tried hard to run RVC/W-Okada with that GPU. Any NVIDIA GPU before RTX 10 series can't even run AIs.

[flat zealot]

tried using uvr5 ui yet got an error, not sure how to fix it or what it even means tbf lmao

Traceback (most recent call last):
File "D:\UVR5-UI\env\lib\site-packages\gradio\queueing.py", line 625, in process_events
response = await route_utils.call_process_api(
File "D:\UVR5-UI\env\lib\site-packages\gradio\route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
File "D:\UVR5-UI\env\lib\site-packages\gradio\blocks.py", line 2103, in process_api
result = await self.call_function(
File "D:\UVR5-UI\env\lib\site-packages\gradio\blocks.py", line 1650, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
File "D:\UVR5-UI\env\lib\site-packages\anyio\to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "D:\UVR5-UI\env\lib\site-packages\anyio_backends_asyncio.py", line 2461, in run_sync_in_worker_thread
return await future
File "D:\UVR5-UI\env\lib\site-packages\anyio_backends_asyncio.py", line 962, in run
result = context.run(func, *args)
File "D:\UVR5-UI\env\lib\site-packages\gradio\utils.py", line 890, in wrapper
response = f(*args, **kwargs)
File "D:\UVR5-UI\env\lib\site-packages\gradio\utils.py", line 890, in wrapper
response = f(*args, **kwargs)
File "D:\UVR5-UI\assets\presence\discord_presence.py", line 146, in wrapper
result = func(*args, **kwargs)
File "D:\UVR5-UI\app.py", line 356, in roformer_separator
raise RuntimeError(f"Roformer separation failed: {e}") from e
RuntimeError: Roformer separation failed: 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_DSA to enable device-side assertions.

gpu is rtx 5080

[low shard]

gtx 900 can iirc
also it's gtx before the 20 serie

[low shard]

are you using eddy's uvr ui?

[hallow thistle]

I mistyped the GTX.

[low shard]

@viscid moss did you add support for the 50 serie in your uvr ui

[flat zealot]

yep

[low shard]

I asked if the creator added 50 serie support in the project, let's see what they say

[flat zealot]

alright

[low shard]

also be sure your drivers and window versions are updated

[flat zealot]

do studio drivers work or do i need the game ready ones

[low shard]

game ready should b fine

[flat zealot]

ok, thx

[viscid moss]

Damn, nope

[low shard]

nice

@flat zealot the following method works for Applio, maybe try it also for eddy's UVR UI:

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 --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128

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

[flat zealot]

worked, thx

[low shard]

you're welcome

[viscid moss]

lmao i was writing the most complete guide

😭

Well, imma pin that as an issue on GitHub, thanks for the pointer Nick

[low shard]

Yw!

[fervent frost]

btw it somehow worked with my i5 4460 cpu but weird 😭 (i selected cpu version and it somehow worked and it didnt give error like my gpu)

[low shard]

yeah as I said it does work, just that the performance is shitty

and could lead to damaging if done for a long time

are u fr using ur cpu? how much high is ur perf

[hallow thistle]

Even worse. This Intel CPU is too old, judging by that Intel Core generation number. While running W-Okada with only CPU is possible, it's usually slower than a GPU. I mean, it must be really slow.

[fervent frost]

na na i stopped using it

[hallow thistle]

Run W-Okada on a cloud service like Kaggle instead, until you get a better new PC with better GPU.

[gentle tide]

whats that error on W-Okada ?

Torch not compiled with CUDA enabled
vcclient_dev\voice_changer\voice_change_manager\models\rmvpe\rmvpe.py:318: FutureWarning: You are using torch.load with weights_only=False (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for weights_only will be flipped to True. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via torch.serialization.add_safe_globals. We recommend you start setting weights_only=True for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
Torch not compiled with CUDA enabled
vcclient_dev\voice_changer\voice_change_manager\inferencer\rvc_inverencer_v2_f0.py:22: FutureWarning: You are using torch.load with weights_only=False (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for weights_only will be flipped to True. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via torch.serialization.add_safe_globals. We recommend you start setting weights_only=True for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
torch\nn\utils\weight_norm.py:134: FutureWarning: torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.
WeightNorm.apply(module, name, dim)

[hallow thistle]

What is your PC GPU?

[gentle tide]

AMD RADEON

[hallow thistle]

AMD Radeon RX or with no RX in its name?

[gentle tide]

rx yes

[simple ore]

local installs for 5000 series require a nightly torch install

[simple ore]

you're using outdated software with new pytorch 2.4+

[gentle tide]

how to fix?

[simple ore]

use forked voice changer

https://rentry.co/forkvoicechangerguide

[hallow thistle]

If you installed W-Okada from start, not from the compiled one, then pretty much it happens.

[simple ore]

it breaks applio's training though, anything after mar 27th is broken

have to specify exact versions

[low shard]

which? elaborate

[low shard]

did you read https://rentry.co/forkvoicechangerguide

never use video tutorials for wokada

[low shard]

perfect

@simple ore since you're a dev for more than a decade, how does it feel dealing with software that breaks easily

[simple ore]

python is worse than Win95 dll hell

have to be super careful at managing the versions and always use a virtual environment

[narrow crypt]

helloo! I was wondering if there's a free open source way to translate speech in videos to another language, making it sound relatively natural and not like a TTS voice-over. And if it is possible, could someone point me to the right direction?

[simple ore]

I've seen Merlin Clone doing good job, but it is paid stuff

[narrow crypt]

For example, I have a video where I'm speaking French and I want to replace the speech to english, but retaining my voice, inflections, etc.

[low shard]

would u say it's the most fragile language u ever tried

the irony is that it's also considered the easiest

[narrow crypt]

oh :T

[low shard]

maybe https://github.com/R3gm/SoniTranslate could help

i'm not sure how good is the quality

[narrow crypt]

I'll check that out, much thanksss!

[low shard]

yw

[simple ore]

https://www.youtube.com/@merlin_clone/videos

[narrow crypt]

another question:
do you guys know about an AI tool that can generate subtitles from a video?

I mean, a free, open source reliable AI tool... 😬

[low shard]

https://huggingface.co/spaces/Nick088/Fast-Subtitle-Maker this is my open source project which uses Groq's free api for a fast whisper usage

I'm not sure if you want to do it locally

[narrow crypt]

to me, the link goes to a 404 page 🥲

[low shard]

I edited the link, there was a typo before

[simple ore]

ASR can be done with Whisper, but it may require some editing and fixing

it is not perfect

[narrow crypt]

locally would be great. but I think that would require my machine to be capable of running an LLM, correct?

[low shard]

for STT, you'd need a ASR model not LLM

what's your pc gpu

[narrow crypt]

right now I only have a 4070 🫠

[low shard]

yeah there's always a little imperfection, just like how LLMs can hallucinate

[narrow crypt]

that's good to know!

[low shard]

I haven't tried on my own machine, but iirc it should be able to handle it

[narrow crypt]

but if it puts the timestamps at the right place, that alone would already be such an improvement to my work

[low shard]

maybe https://github.com/m1guelpf/auto-subtitle ? I'm sorry but I never tried locally

[narrow crypt]

oh, no need for apologies!
it already help me a lot just by pointing me towards these projects

next week I'll try to setup everything and make some tests with the translation and subtitles

muuuch thanks ✨ 🙌

[low shard]

goodluck and ur welcome

[vale osprey]

-colab

[patent trellis]

📒 Google Colab Notebooks

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

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

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

by Jarredou & Makidanye
Google Colab

[brittle wing]

@simple ore

[slender osprey]

I followed the guidelines in the manual to reduce the delay, but I can't hear my voice modified here, how do you hear? I tried to watch the tutorials here but I still found it confusing because their audio mode is as "client" and mine is as "server" (AI Voice Changer)

[brittle wing]

That's applio Colab also wdym

How was it done

Huh

That's Applio Colab with the fixes

[simple ore]

read it again

[brittle wing]

It used to work fine w the fixes...

[simple ore]

wrong spelling

so I dont know what the hell you're running

code's been updated yesterday, it does not have misspellings

[brittle wing]

Found it...

[simple ore]

wrong code

not up to date

this is current

https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio.ipynb

[ancient ocean]

There is the bug in Applio UI in Colab when traning:

[simple ore]

UI or noUI?

[ancient ocean]

UI

[simple ore]

does the install applio cell look like this: #✨│ai-help message

exactly

[ancient ocean]

The previous bug related to gradio successfuly fixed by myself after adding --prelease disallow and !mkdir entities to beging of Install section. Now it gone worse )

[ancient ocean]

Currently:

@title Install Applio

import sys
from IPython.display import clear_output
import codecs
import os

encoded_url = "uggcf://tvguho.pbz/VNUvfcnab/Nccyvb/"
decoded_url = codecs.decode(encoded_url, "rot_13")

repo_name_encoded = "Nccyvb"
…!uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1
--index-url https://download.pytorch.org/whl/cu121
-q --prerelease if-necessary-or-explicit
!uv pip install numpy==1.23.5 gradio==5.23.1 pyngrok -q --prerelease if-necessary-or-explicit

clear_output()
print("✅ Finished installing requirements!")

[simple ore]

what is

[ancient ocean]

copy paste issue

[simple ore]

change numpy version to 1.26.4

[ancient ocean]

Another detail: They should change program_ml to Applio in dir paths of Backup Sections:

"/content/program_ml/logs/" should be "/content/Applio/logs/"

[ancient ocean]

It works!! Thank you

[simple ore]

okay, gonna update everything again

[low shard]

U gonna have a streak of updating colab 😭

[rancid crown]

hi in here?

[low shard]

#✦│chat message hello

Your rtx 3070 is good enough

The issue is all video tuts are old

Delete everything you got off YouTube tuts

If they made you install vb audio cable, uninstall it too as it gives you random issues on windows

[rancid crown]

yh they did

is it that bad

[low shard]

YouTube tutorials use the old version of the original Wokada, which has worse performance and quality

[low shard]

Yes, it could just randomly stop working out of no where

[rancid crown]

that might be why

[low shard]

It's not stable on windows as users reported

-realtime

[rancid crown]

shall i delte everything even the luacner

[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]

Delete everything you got off the YouTube tutorial, then read the 1st guide

[low shard]

Yeah delete everything you got off youtube

[rancid crown]

okay ill do it now

where can i find the guide

[low shard]

The 1st link of the bot message above

[rancid crown]

damn looks really complicatewd lol

[low shard]

Nah it's just because it got everything for all OS and all GPU types, you need to only follow the general parts like the 3rd part and the audio setup, and the windows Nvidia parts

Ofcourse if you got any issues you can ask here

[rancid crown]

so its not as eay as just donloading it

[golden walrus]

guys, can i ask if it's better to have long data ? like 2-3 hours

[low shard]

it's easy compared to the old original wokada where you needed to install python and other shi

[rancid crown]

gonnaaa be honest no idea what im doing with with the steps

so lost

[low shard]

1h is enough

[low shard]

it's AI, it's not going to be like downloading minecraft since this is supposed to be advanced btw

it's as easy as downloading 2 programs, extract, and run them

that's kinda all

[low shard]

what step are you at?

what is going wrong ?

[rancid crown]

i struggle for minecraft ahaha

[golden walrus]

so is there anyway to train it and make it sound natural in my mother tongue ?

[rancid crown]

nevr been good with this stuff desoite gaving a computer for 3 years

always asking everyone for simple help lol

[rancid crown]

the lot

[golden walrus]

i tried with 1 hour and the result still sound like some British come to my country

[low shard]

you're cooked then

🔥

/ jk lol

[low shard]

what's your mother language

[low shard]

the lot ?

[golden walrus]

Vietnamese

[rancid crown]

ik wont be able to do it sooooooooooo

accepting my fate

[golden walrus]

i know it's not possible to train my own embedded model

[low shard]

RVC original pretrain is trained in english only, it's also good for languages similar to english

you could try and check the tensorboard, it may be better if there's a vietnamese pretrain

[low shard]

I can help you up!

what step don't you understand?

[rancid crown]

lemme properly look at 1 causde i looked and it scared me so i close it ahah

[golden walrus]

so i should find a Vietnamese pretrain right ?

[low shard]

if there is one, it could help yeah

[golden walrus]

if there isn't any, is it possible to train one ?

also, embedded model is different from pretrain right ?

too low tech for these

[simple ore]

embedder model is the one that extracts phonemes from the audio

[rancid crown]

i can try late ris thats good kinda busy rn

[golden walrus]

and thank you for clearing my confusion

[low shard]

alr

[golden walrus]

oh okay, i get it. thanksssssssssssssssssssssssssss

[simple ore]

i dont know vietnamese embedder

but if you use one to train a model then you can only infer using the same embedder

[crystal rivet]

I have a nvidia geforce gtx 1660 super
Is it still possible to get Okada's Live Voice Changer to work?

[low shard]

yes except for high games

also be sure to NEVER USE VIDEO TUTORIALS, they are all old for wokada

if you used one, delete everything you got off it

[crystal rivet]

I see.. thanks

[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, wokada deiteris fork

this is the most up to date guide

the deiteris fork has way betterr performance and quality compared to the old original wokada used in the yt tuts

@crystal rivet read it up and let me know for any issues!

[crystal rivet]

Is there a percentage of used gpu I should watch out for while running games?

[simple ore]

you can try with this one https://huggingface.co/utter-project/mHuBERT-147

[restive granite]

hi

[low shard]

wdym?

[restive granite]

i need testers

[low shard]

it's normal that your gpu is going to be at an high percentage

[simple ore]

unfortunately there's no control for gpu usage or prioritizing one app over another

so a game would be competing against the voice changer for resources

[restive granite]

i create AI on the symulator of predyctions the lotto meybe lottery is not good point but this machne learning even 1000 eproch

[low shard]

you can try #1359898289335566570

[golden walrus]

yes sir

[restive granite]

what now? you can promote this on main options then to join to chanell he must vistit me website? XD i can pay for it 😛

[upbeat sparrow]

If you can predict the lotery why aren't you cashing the prize yourself lol

[upbeat sparrow]

Can i record a tutorial then ?
I only knew of the original/old one before joining here,the fork is way less famous i assume

[crude flame]

go ahead we need a yt tut

just make sure you suggest the correct settings and such

[rich marsh]

guys i need a tts on colab cuz trash pc i have. i tried to find but they dont wrok :(((8

[golden walrus]

sir, can i ask if there is a tutorial for installing custom embedder on applio, i tried to download mHubert but i have no clue how to do it properly. and my applio failed to find it

[simple ore]

you need pytorch_model.bin and config

[golden walrus]

i have this

but still

[golden walrus]

it only use contentvec

[simple ore]

one sec

[golden walrus]

really appriciate your time

[simple ore]

save two files

click (x) custom, click Refresh Embedders

[golden walrus]

okay, let me do it

[low shard]

There are different Text To Speech (TTS) AIs:

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

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

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

You can check TTS in our tts index

With RVC Models:

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

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

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

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

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

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

• You could try another tts from our tts index and use the output as an input in rvc

[low shard]

yeah u can make a yt tut

[crude cypress]

whats the best realtime voice changer that works good for AMD

Cus w-okada is laggy for me

[steel forge]

-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

[steel forge]

First link on there

Modified deiteris w-okada, one of the big features is better AMD support

[crude cypress]

https://github.com/deiteris/voice-changer/releases/tag/b2332

Which

[steel forge]

Extremely loud incorrect buzzer ERRRRRRRRRRRR

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

This section in that guide

Its a direct download

[crude cypress]

I think i found it

Its the same file right?

[steel forge]

dml-blahblah.zip

[crude cypress]

Yea alr

[steel forge]

DML is what you want

[crude cypress]

Cus i have AMD?

[steel forge]

Ye

[crude cypress]

would it work smooth with a rx 7800xt and a ryzen 7 7800x3d

[steel forge]

Oh yeah that's like too end shit

top*

You'll be good

[crude cypress]

Alr

[steel forge]

That guide has good recommended settings and stuff. Its a really really good guide, tons of info

Also AI Hub search bar is way more useful than Google for troubleshooting this

[crude cypress]

Where is the AI Hub search bar

Sorry im blind

[frosty valley]

yo can anyone help whats the best voice changer for gtx 1650 and ryzen 5 4650g pro?

[terse spade]

Why can't I use the applio black interface ?
Can't download output audio
Still stock on 0:00

[gloomy kernel]

Hi there! I'm new to AI voices, Using RVC at the moment. I want to learn how to make a voice for a youtuber but I'm not sure how. Any advice?

[knotty moth]

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[knotty moth]

https://rentry.co/forkvoicechangerguide

[hallow thistle]

It's W-Okada. The free and open source realtime voice changer program.

[azure oxide]

iiii

[mint goblet]

currently trying to train a model via mainline but im getting this issue

ERROR:  
ERROR:  Sign up for an account: https://dashboard.ngrok.com/signup
ERROR:  Install your authtoken: https://dashboard.ngrok.com/get-started/your-authtoken
ERROR:  
ERROR:  ERR_NGROK_4018
ERROR:  https://ngrok.com/docs/errors/err_ngrok_4018```

i added my ngrok token to the secrets tab + verified my ngrok account (the credit card thing) but im still getting this issue

[knotty moth]

• make sure you've entered your correct authtoken from https://dashboard.ngrok.com/get-started/your-authtoken
• make sure you haven't hit any rate limit as shown in https://dashboard.ngrok.com/usage
• no, you don't even need a credit card to use it, unless you want to lift the rate limit

[mint goblet]

wait

oh shoot i misread the error lmao

i thought the verified account thing was like the identity verification thing

[knotty moth]

it's just email verification and no more hassle for me

[mint goblet]

do i need to do the ngrok config add-authtoken $YOUR_AUTHTOKEN thing
and if so where do i do that? havent used kaggle + dont see a cmd line or anything

[knotty moth]

ngrok config add-authtoken 123456abcdef
(paste from the one in https://dashboard.ngrok.com/get-started/your-authtoken)
do you still not get it?

also no, I shouldn't give my token to anyone

[mint goblet]

sry i think i messed up my phrasing
on the kaggle notebook it has a place that says
``` #@markdown To put ngrok token, go to the side and press the key icon<br> On the name put Ngrok_Token and the Value to be your ngrok token

i put my ngrok token in the secrets tab (key icon under add-ons tab) as they asked in the instructions

i was wonderin if i still needed to do the ngrok config thing on my own on top of that or if adding the ngrok token to secrets was enough

wait nvm i found the issue

my dumb ahh added a b when i pasted in my token

[knotty moth]

also no, I shouldn't give my token to anyone

are u still cooked?

[mint goblet]

it works now

[knotty moth]

ok then

[mint goblet]

https://tenor.com/view/peko-wink-gif-23946907

[knotty moth]

https://tenor.com/view/ninomae-inanis-ina-embed-fail-embed-fail-meme-holo-live-gif-24386703

[mint goblet]

pain peko

[final jacinth]

Hey, can someone help me? I recently reset my computer, and thus, reset all permissions (unintentionally), so now I have to run everything as admin, if I don't, it gives this error (image). However, when I run start_http.bat as admin, it immediately opens up a small window, closing it within the second. I don't see anything in task manager either, is there some way to fix this (preferably not changing Windows settings)?

Uh, for W-Okada.

[knotty moth]

there is a space in W-Okada New

[mint goblet]

btw is the rate limit monthly?

just hit my limit (managed to finish training + downloading my model first tho thankfully)

[knotty moth]

you'll hit the rate limit if you download more than 1 GB (reset every first day of month) in the file manager instead of kaggle's right sidebar

or the HTTPS Endpoint Requests reaches limit

[final jacinth]

It still refuses to work.

[mint goblet]

gotcha, tysm

[knotty moth]

move to another place like D:
and check the directory permissions

if there's only a C: drive, I'd suggest creating a partition depending on the free space or get another external/USB drive

[final jacinth]

Well, it's not sending a traceback to me, so something's happening.

Oh yay it worked.

Thank you very much.

Now I know that my D:\ actually has permissions for shit lol.

[weak forum]

im new to all of this in the video i watched for so vits fork it says i need a config file which should be in the form of a JSON file but every single audio model i download there is a pth which is fine but the other one is always an INDEX file. Am i doing something wrong?

[low shard]

So vits SVC is old ass

All YouTube tuts are old

https://discord.com/channels/1159260121998827560/1361235120677195877 let's talk here

[low shard]

What's your PC GPU and be sure to not use yt tuts

[low shard]

Realtime for calls? Wokada deiteris fork

-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

[frosty valley]

alr bro thanks

[low shard]

Let me know for any issues

[frosty valley]

alr bro

oh yea check this out

[low shard]

?

[frosty valley]

is say's compressed zipped file

do i have to extract it?

i just like downloaded the voice changer windows nvidia-b2332

[low shard]

Yes

[low shard]

Yeah it's for windows nvidia

[frosty valley]

alr so im already on this page: "If nothing opens after a while of codes loading in, then open a browser and type in http://127.0.0.1:18888/. This is a local URL, it runs on the WebUI."

what does it mean?

[low shard]

Did you open the .exe first of all

[frosty valley]

idk? but i open the mmvcserverSIO like the application the guide told me to open to

[wind scroll]

People would like some help, I have an iPhone, can I use some aihub panel or sla to change a person's voice to mine using the iPhone?

[low shard]

Yes that's the .exe

Did it open your own browser

Show a screenshot

[frosty valley]

yea

[low shard]

Yeah that's normal

What's your browser btw

[frosty valley]

opera gx

[wind scroll]

please help me with this

[low shard]

Ai hub panel? Ai hub isn't a program

Are you trying to change your voice in realtime for calls, or just prerecord audios?

[wind scroll]

prerecord audios

[frosty valley]

so uh what settings do i pick? cause like idk what the chunk, extra, fo det? like whats the best settings for gtx 1650?

[low shard]

Users reported opera gx giving issues for wokada, also that browsers fancy effects can cause performance issues bc it eats ram, personally we don't suggest it,

We would suggest something like Chrome or Firefox

After u do that, show a screenshot of ur wokada

[frosty valley]

oh ok ok

[wind scroll]

Is there a panel or something I can use to upload the original audio and my voice to swap?

[low shard]

There isn't a panel

Ai hub isn't a program

[wind scroll]

thanks

[low shard]

RVC is the program that does speech to speech conversion

I gave you links to do it on cloud since mobiles aren't that much good to running it locally on the hardware

[low shard]

Yw and lmk

[wind scroll]

I didn't understand very well, I'm using the translator, I've already separated the audio from the background music, and I've already separated my own audio, where would I go to change the voices? I use an iPhone

[low shard]

You should use one of the sites i sent you

a mobile phone isn’t powerful enough to do it locally (running on the phone power’, so you need to use cloud (remote good pc with limited free time)

[violet light]

Hey! Hope you're doing great. I build custom GPT chatbots for business owners — they handle DMs, answer questions, and help customers 24/7 without you lifting a finger.

It’s like having a smart assistant that works nonstop. Want a quick preview?

[kind laurel]

How does one make an ai rap song with like drake I have the lyrics but idk how it all works

[low shard]

so make your own song with modified lyrics?

[wind scroll]

How do I turn my audio into a template to use in RVC?

[knotty moth]

https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[verbal oasis]

hi, what is the easiest method/tool to use to convert an existing voice from an existing audio to another voice?

basically speech to speech

is there a colab or something?

[low shard]

RVC, what’s ur pc gpu

[hallow thistle]

It's RVC.

I remember you talking about this somewhere. You never told me about your PC GPU but implying your PC won't gonna run Applio that well.

[steady field]

guys i got a voice model now how do i use it

[low shard]

What's your PC GPU?

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

Do you need Wokada or RVC?

[timid bone]

on the voice changer I cant see any output devices

[eternal oasis]

Hello, since I updated Torch for compatibility with RTX 5000 cards I have the feeling that in applio the voice conversion adds an English accent to my voices whereas I did not have this problem before. Anyone else encounter this problem?

[simple ore]

torch has nothing to do with it

[eternal oasis]

ok then i think is about settings in the new applio version

[simple ore]

use of index brings the accent of the model

french audio + english model + 1.0 index may result in english accent, 0.0 index probably not

[eternal oasis]

ok i will try that thank you

[low shard]

Elaborate

What's your PC GPU

What browser are you using

What tutorial link are you using

What you want to do

[hallow thistle]

There are programs that use the RVC voice model:

1. a regular voice changer program that converts audio file like Applio
2. a realtime voice changer program like W-Okada
   Which one are you looking for?

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

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

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

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

__**BE PRODUCTIVE.**__ 🤝

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

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

[hallow thistle]

Which W-Okada version do you use? What is your PC GPU? And have you installed a microphone or virtual line in program?

I expect people to respond at least within a few minutes. If they don't respond for a very long time, they must be busy like crazy.

[glacial finch]

what settings is recommended for the client

is there like a global rommendation?

[pastel oak]

whats your gpu and send a screnshot of your voice changer

[potent bone]

where can i get the newest version of w-okada? my gpu is 3060 ti

[glacial finch]

like theres also a tab called advanced settings i dont wanna play with it god knows what will happen

[analog obsidian]

change f0 to rmvpe
disable sup1
enable fp32 in advanced settings

[glacial finch]

idk if i need to change things there

[glacial finch]

thanks for the help 🙏

[analog obsidian]

after enabling fp32, restart the voice changer
wokada has a bug that everytime the precision is changed, it'll use more resources for no reason at all

closing and opening the voice changer fixes it

[hearty peak]

!help

[tawdry gull]

LunaBot 🌙#9997

LunaBot 🌙 is the perfect music bot! Feature rich with high quality music! And Custom Playlist

You can start listening music by just joinning a voice channel and typing: /play [song name or link] (Remove brackets).
We support only Spotify, soundcloud, bandcamp and more!

To view more help on a specific command or category, run
/help <command> or /help <category>

Important Links:
Support
Premium
Invite

Command Categories:
🎶: Music
💰: Premium
⚙️: Utility
📕: Admin

Select A Page From Dropdown Menu Below

LunaBot · Apr 14, 2025, 4:42 PM

[glacial briar]

​No Category:
  help Shows this message

Type !help command for more info on a command.
You can also type !help category for more info on a category.

[hearty peak]

where to find w okada voice changer tutorial?

[analog obsidian]

https://rentry.co/forkvoicechangerguide

[limpid plover]

trying to use the new RVC AI Cover Maker by @viscid moss and I've got no idea what's going on with it. separation works fine, but the actual inference fucks up. tried both the precompiled version and doing the install from scratch and it just won't work!! How do I send the traceback? It won't fit in just one discord message lol.

something about a yaml ConstructorError

[viscid moss]

Local installation is broken, just cloud for now

[limpid plover]

aah so desu ka

well thanks for replying, saves me further headaches!

[stoic storm]

yo regarding RVC

Im trying to use Applio

and i need to insert the Model link in the public link. How do i get that link if i use google drive?

[bold valley]

Where I could share my ai song in AI Hub server?

[low shard]

promos ain't allowed

[low shard]

download the model, extract it on ur pc, upload the model manually in the download tab of the applio colab public link

[finite parcel]

Hi
I would like to ask for your opinions on which AI platform: GPT, Grok, or DeepSeek would be better suited for tasks such as homework, research, and similar activities. Or any other AI i didnt mention that is better
The one i most use is GPT, but when the daily paid trial of GPT is unavailable, which alternative would you recommend?

[cinder sable]

can u guys tell me whats settings are the best for the voicechanger. i cant send screenshot my gpu is 4070

and is there a way to lower the delay and make the sound less robotic

[low shard]

tbh i personally prefer gemini 2.5 pro from aistudio.google.com

for a sec i thought this was ai chat lol

[low shard]

!give-media-perms 1h @cinder sable

please send a screenshot

[brittle wing]

How do i download the voice changer owkada

[low shard]

#🧬│ai-chat message since you got a gtx 1650

i think you want realtime voice changer for calls/games

-realtime

[brittle wing]

yeah

[low shard]

you need to read the 1st link, wokada deiteris fork (modified version) which has better performance and quality

ayo

where tf is the bot

[brittle wing]

i got just got that rn

[low shard]

https://rentry.co/forkvoicechangerguide

[brittle wing]

have everything open rn

[low shard]

it's this guide

[brittle wing]

got it

[low shard]

nice

[brittle wing]

yeah

[low shard]

show a screenshot of ur wokada i can help with settings

[low shard]

damn ur using an old version of original wokada

never use yt tuts for wokada bro

delete that along with vb cable

[cinder sable]

i didnt know its my first time using this bro

[brittle wing]

cant send screenshots

nere

[low shard]

don't worry, we are there for help ofc

[cinder sable]

soo what should i download?

[low shard]

wokada deiteris fork: https://rentry.co/forkvoicechangerguide

[cinder sable]

okay ill be back

[brittle wing]

what do i needa configure?

[low shard]

uncheck sup1
f0: rmvpe without onnx
extra: 2.0

close useless apps in background

are u using it along games?

[brittle wing]

yeawh

[brittle wing]

why is there 2 extras?

[low shard]

wdym?

[brittle wing]

also its a bit buggy so should i higher the chunk

[brittle wing]

you said

extra: 1.0
extra: 2.0

k

[low shard]

lmfao my bad brotha, use the 2nd one i said

[brittle wing]

k

what chunk?

[low shard]

higher the chunk to a value a bit higher than perf number

also, are you using it with games? if so lower the graphics to the minimum

[brittle wing]

yeah i did

its on preformace mode

[brittle wing]

what chunk u suggest?

[low shard]

u can also lower the extra more which will give u less delay but cost u some quality

[low shard]

there isn't a perfect value, it controls the delay, you just need it like perf value + 60 for example

[brittle wing]

i want good quality

its supper buggy

like super

[low shard]

alright keep it as it is then
you could also try using https://rentry.co/forkvoicechangerguide#reduce-more-delay for less delay (wont affect quality, just isn't a 1 click thingy to do)

[low shard]

wdym? can u show ur wokada again and elaborate

[brittle wing]

RN

[low shard]

uncheck only sup1

also, what issue are you facing now?

[brittle wing]

hmm

wait

hmm its tiny laggy

the voice

[low shard]

are you saying the voice sounds robotic? or another issue

[brittle wing]

kinda

just a bit

[low shard]

u can try to higher up extra to 2.7 for a bit more quality but it's going to have an higher delay

tbh it would be just better you try another model

not all models are super high quality, there's over thousands

[brittle wing]

do u have one i can try?

[low shard]

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

[brittle wing]

I'm using duckus Egirl rn

[low shard]

alright

[brittle wing]

should i higher the chunk a bit more?

[low shard]

u should do that only if ur perf value is higher than chunk

[brittle wing]

wym perf value

[low shard]

u will see the perf value when u click start

[low shard]

gpu: rtx 4070
f0: rmvpe without onnx
extra: 2.7
chunk: 128 ms (this is generally, to find ur own best chunk just put the chunk higher than the perf value a bit)

[cinder sable]

is 50 gonna be alr?

[brittle wing]

i think i got it to work with a differnet model

its nice rn

760ms

2.7

[low shard]

dw

so is there any other issues or is it solved

[low shard]

nice, u could also try the link i said before to have less delay if u wanna

are u facing any ther issue or is this solved

[cinder sable]

can i like make the sound more realistic or its gonna be how it is do i need to try some things out

[low shard]

u can try playing with the pitch or use other modelsd

[cinder sable]

okay so if the pitch doesnt help i should try other model. no need to change other thing right?

[low shard]

U could try the index too but generally just try a better model

[cinder sable]

okay ty

[eternal steppe]

can someone share there settings

help

pwease

<@&1159293204038955078>

my shit sounds ass bruu

i need sum sttings

[viscid moss]

https://rentry.co/ForkVoiceChangerGuide

[eternal steppe]

i cant hear myself but when i record i can hear the audio

[eternal steppe]

<@&1159293204038955078>

<@&1159293140440723499>

[viscid moss]

Pls stop pinging roles

[eternal steppe]

mb

but can u help

i cant hear myself

but when i tap th e record button i and finish

i hear the voicechanger

[viscid moss]

can u send a screenshoot?

[eternal steppe]

how? like can i explain it in a screenshot

am tryna say is

ik the voicehcnager is working but i cant hear myself with it

but when i rec and tap start then stop i can hear the input and output audios and it works perfectly fine

??

[viscid moss]

on "monitor" u can set an output device to hear yourself

[eternal steppe]

ohhhh

apriecate it

[knotty moth]

(have turned off role ping notifications already lmao)

[knotty moth]

you should only ping mods to report any inappropriate behaviors

[hallow thistle]

Also, please don't be down bad pinging mods just for this especially. Not all mods know about W-Okada.

Don't be hurry, calm down, ssit down and listen.

Ok, let me start this. What is your PC GPU? Which W-Okada version are you using? Because most of the time, the "I can't hear myself" phrase always happens to those who followed tutorial videos on YouTube which only telling you to install the original version of W-Okada and VB-Cable. VB-Cable gives random issues to Windows users.

[sour sun]

Some input on this: On the fusing tests (KLM + OG) I did, the resulting models had new strengths and weaknesses

Strengths: much more sensitive to input audio, was "softer" and seemed to generalize better?

Weaknesses: sounded much less like the voice overall

[steel forge]

Honestly think vbcable is a skill issue

I know a lot of people have problems with it but I have never seen it

The other cable on the guide is prob better anyway

My favorite is the cable that says TRIAL every 30 seconds

[knotty moth]

most video tutorials keep recommending it

[knotty moth]

that's when ppl download through the website (and get jebaited by the trial version) instead of the VAC lite direct download

[low shard]

Have patience, don't ping multiple times

[rare bramble]

is refine gan available on main branch or codename's fork rn?

[stray steeple]

any1 have the tensor board latest update link so I can install?

[steel forge]

Its included with Applio

https://github.com/IAHispano/Applio/releases/tag/3.2.8-bugfix

This is the latest applio

[hallow thistle]

Applio has Tensorboard built in. If you double click and run this batch file, it will launch its web UI in your browser.

[void holly]

every model apart from the jvs stock one crackles so bad that my voice in inaudible

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

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

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

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

__**BE PRODUCTIVE.**__ 🤝

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

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

[hallow thistle]

What are you looking for? RVC or W-Okada or the voice model?

[void holly]

i'm using w-okada and im trying to figure out if its a problem with on my side or with the voice models

[hallow thistle]

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

Audio sounding robotic can happen when you didn't set chunk and extra settings right, or it can be the voice model itself.

[void holly]

2.0.76 beta aml rtx 3080

[hallow thistle]

Never use the DirectML version with NVIDIA GPU, especially the original one which is outdated. Download and use the better W-Okada from there instead. https://rentry.co/ForkVoiceChangerGuide#download-nvidia-on-windows

[void holly]

thanks ill give it a try

is there any reccomended settings for my gpu

[hallow thistle]

W-Okada DirectML is made for AMD/Intel GPUs. While it is possible to use that variant of the program with NVIDIA GPU, it will give the slightly worse performace than that full NVIDIA-variant W-Okada.

[void holly]

alright thanks

[void holly]

the chunks and extra

[hallow thistle]

Read some more on the guide I sent to you. https://rentry.co/ForkVoiceChangerGuide#for-nvidia-on-windows

[unreal hamlet]

Hi, any idea why its not working on my PC? These are the logs:
https://mclo.gs/uGnEVRD

[hallow thistle]

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

[low shard]

lemme guess, you followed a youtube tutorial? You seem to be using original wokada

[hallow thistle]

It's crazy. Using the original W-Okada in 2025.

[unreal hamlet]

vcclient_win_cuda_2.0.76-beta.zip, 5070

[unreal hamlet]

a

[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]

read the 1st link

the wokada deiteris fork

the original wokada doesnt support the 50 serie

the deiteris fork is more optimized and supports it

[hallow thistle]

I knew it you used the original version of W-Okada lmao. But you also got NVIDIA GeForce RTX 5070.

[low shard]

if you followed yt uts, those are old, uninstsall everything you got off youtube tut wokada

[unreal hamlet]

I just remembered that I downloaded it once and looked for it again

[low shard]

it’s better you follow only the written guides

follow the 1st link i sent

those are updated

meanwh ile yt tuts are pretty old

[hallow thistle]

The original version of W-Okada is outdated, and it won't gonna work with RTX 50 series GPUs. Original and Detris' W-Okada are two different versions of W-Okada made by different authors.

[low shard]

[steel forge]

Namari spitting straight facts like usual 🔥

[low shard]

namari on 🔥

[unreal hamlet]

thanks guys, the best!

[low shard]

let us know for further issues or questions

also a reminder to update drivers and windows version in case of issues

@unreal hamlet btw sorry to disturb but hows the 50 serie experience going? Still having some issues like black screen sometimes or missing rops?

[unreal hamlet]

ty

[unreal hamlet]

In my experience, it has worked well for me so far, I have not encountered any problems or visual failures, it simply works as it should hehe

[low shard]

did you check if you had any missing rops as well?

some unlucky people had encountered at unfortunately

[unreal hamlet]

dont check that :c

[low shard]

if u haven’t checked and want to check https://www.pugetsystems.com/blog/2025/02/26/nvidia-geforce-rtx-50-series-missing-rops-issue/ could be helpful

or idk if u were saying it in a way ur too scared to check bc u know some rops are missing 😭

[unreal hamlet]

I think it s okay...
https://gpuz.techpowerup.com/25/04/15/cmr.png

[jade orchid]

im using a 3080 am i cooked?

[low shard]

yup u are all good dw

[low shard]

what do u want to do?

An rtx 3080 is not bad at all

[astral shell]

ive been so confused, ive had a meltdown with chat gpt for the past 4 hours trying to install RVC and I just cant get it to work, it countinuesly gives me errors left and right when I accesed the panel and tried converting my audio, can any1 help?

[low shard]

youtube tutorials are old and chatgpt doesn’t know about rvc well

wha’s your pc gpu and what do you want to do

[astral shell]

wanna train own pop smoke voice model, and also convert other acapellas so that it works n all, 4070 super

[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

i suggest applio personally

[astral shell]

whatever doesnt bring me life regrets works and whatever doesnt take a whole decade to install

[low shard]

its easy as downloading a zip and extracting it

[astral shell]

Sweet, but ive installed a whole bunch of python stuff, should i delete python fully and reinstall or keep it?

[low shard]

python is embedded in the program

you don’t need it system wide

[astral shell]

so i can just delete python now ye?

[low shard]

If u want to yeah

[astral shell]

aight bet

[low shard]

it’slike u have 2 different pyhtons

one is the one installed system wide

one is inside the program files

[astral shell]

Ye fair enough

python taking hella space

[astral shell]

do i need any of thisd? cause i needed this when using mangio rvcv

ah cant send images

do i need microsoft build tools? (some c++ program which takes like 10gb sapce)

[low shard]

Mangio rvc is abandoned since 2023 lol

you can delete everything you got off youtube and chatgpt

[astral shell]

aight bet

[low shard]

hows it going

[raven perch]

Hi there. I have a question and this seems like the right place to ask.

[astral shell]

All good currently extracting

[low shard]

what is it?

[raven perch]

Is there anyone here who could analyze videos? Or is there like a detection tool for AI?

I recorded a real; raw video but someone is convinced it’s fake/ AI and I’d like to put that to bed.

[low shard]

about ai detectors, i think they are mostly about text, i saw somewhere there are ones for images, but anywaysh you shouldn’t trust much ai detectors

they aren’t accurate and can lead to false positive easily

I can say this after multiple tests i did personally tbh

like all you need is tell the ai a good prompt and the detector would say its 99% made by a human

Also i’m sorry to hear people are accusing your video of that

[raven perch]

Ok. Is there anyone here who is good/ semi-professional at analyzing videos?

This video is quite important. It contains evidence that is worth quite a lot of money.

[low shard]

well i don’t think there’s actual professional investigators here, tho you could check the frames of the video and point out some details to the person

I think the best way to prove someone that your video isn’t ai is to show how your video was made, like if it’s an animation show the draw sketches or things like that

[raven perch]

I literally recorded the video with an outside phone (so I recorded a conversation on phone A with phone B/2).

I wrote stuff on a piece of paper and had audio noises from the environment to show there were no cuts.

And he still said fake.

[low shard]

wait, is this a case of people accusing a video was made by ai in legal court?

[raven perch]

This is not about a court case.

[astral shell]

Yo so its extracted right, Just bit scared to mess sm up, where do I drop in my ai voice thingy tht Ive downloaded?

[low shard]

you could also point out that usually ai videos look like a slowmotion and that there is some glitches even in the most advanced tools right now

And that most ai tools can only make short videos

like 10 secs or up to 1 minute in some cases

like i haven’t seen any ai yet being able to make a 5 minute or lomger video

[raven perch]

My video is over a minute long; it also shows real-time tweets and I sent out messages to real people during the recording which is seen.

[low shard]

i mean technically everywhere u want, id suggest a dedicated folder ofc

[raven perch]

Would I be allowed to show it to you? I’m not really sure whether this chat is the right place for it cause the information is quite sensitive and don’t know who is lurking.

It’s not NSFW.

[placid wind]

yo quick question, i go with the deiteris voice-changer instead of the orginal one right?

[low shard]

it’s always better to not share sensitive information over the internet, it’s better to not send it in the server

[low shard]

yup

[raven perch]

Would you be ok with a DM?

[placid wind]

alr thanks

[knotty moth]

it shouldn't be affected by those issues on 50-series

[astral shell]

I downloaded the model but how do I use it?

Like I downloaded it on the thing

But after refreshing i still dont see em

i figured out how to use it, but after converting and trying to download it syas "file wasnt available on site"

[restive birch]

I have a question, I'm not sure if it's here I should ask it. When you download a voice model on weight, there's some model where we can change the language of the voice(example the model in in English but there's a french speaking voice too).

How do you know if it takes into account?

[low shard]

lmk

[low shard]

elaborate and show a ss

[low shard]

they don’t like convert the model, the model just used a french audio sample

[restive birch]

Oh I see, well thanks for the answer. Really appreciate it ^^

[low shard]

ehh is this something legal or something personal? I’m not a legal professional lol

[low shard]

you’re welcome

[brittle wing]

yo anyone know how rizz records make his videos if you do plz dm me ive been trying to figure this out for the past 3 months

[astral shell]

I figured it out, now the only thing is that I am trying to install models with only the .pth file, since those are the ones i got but it needs a index file too, possible to bypass somehow?

[hot lagoon]

I have a question for a voice changer kaggle at what point do I run to sell merge the data

[low shard]

You can just not upload the index

[low shard]

Elaborate pls, along with your PC specs, what guide are u using and the issue

[astral shell]

Throws an error without the index

[knotty moth]

use any other index file and set index rate to 0

[astral shell]

got no clue where to set index rate to 0

[low shard]

Can you show a screenshot

Or you could screen record u uploading the model then using it

[low shard]

Advanced settings

[astral shell]

thought i didnt have perms

[knotty moth]

no, the console window

[tame mica]

applio wont infer for some reason

[astral shell]

like this ye?

[knotty moth]

try the 3.2.8 bugfix version if not yet

[astral shell]

ye thats what i tried

[elder coral]

bro i forgot something can you send me something how to make a model using colab?