#✨│ai-help

1 messages · Page 323 of 1

hardy yew
#

Going below chunk size of 300ms will introduce a minor quality drop. Pitch extraction accuracy needs some context and 0.3s is the best compromise here.
Still, lowering it slightly shouldn't have a massive impact. Feel free to try it and see if the output's OK for you

low shard
#

you should set the chunk just a bit more higher than the perf value

#

you could also try vonovox latest stable or beta version

crimson geyser
#

can someone give me the best vc for amd? win11

#

ty

steady turtle
#

Hi everyone, sorry but I just can't find the solution. I would like to change the words of a famous song. The voice and the singing must be the same, I just want to change the words he says. How can I do it? I can't find it anywhere, but I've seen videos of it being made.

hallow thistle
viral mason
#

Why would you prefer dms? Btw I might not be able to help much atm

#

Nothin sketchy I'm hoping

frozen meadow
#

can someone help me download this?

#

im not really good with computers

low shard
low shard
#

!howtoask

patent trellisBOT
# low shard !howtoask
❓ How to Ask for Help
✅ Before You Ask!
  1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
  2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.
📝 How to ask?

Tell your:

  • Full GPU Name: (e.g., NVIDIA RTX 4060 8gb vram desktop)
  • Operating System: (e.g., Windows 11)
  • Detailed Description: What were you trying to do and what went wrong?
  • Tutorial Used: Link to the guide you were following.
  • Screenshot: A picture of the full error message.
🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

  • ANY illegal activities.
  • NSFW/Porn.
    Requests for these topics may be ignored, not helped and result in moderation action.
<:matsuripray:1159685390156967936> Community Expectations
  • Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
  • English Only: Please keep all conversations in English.
  • Don't Ask To Ask.
hallow thistle
river parrot
#

valorant

#

why i cant use it on discord or valorant

low shard
hard osprey
#

I am trying to create a slay the spire anime but having a lot of trouble with basically everything. I think if I learn a workflow and how this clip was made I can edit the work flow and prompts to fit all the other scenes I have in mind. Can someone please create or guide me to be able to make this short clip.

Cinematic fantasy battle scene in a dark swamp environment, fog rolling over shallow water, twisted trees and hanging moss. The camera starts at a medium distance.

Ironclad stands confidently holding his sword. Ten feet in front of him stands entomancer

ironclad smirks mockingly and says: I'll help by hitting myself.

He jokingly punches himself in the chest

1 second pause.

Suddenly ironclad lunges forward through the swamp water. As he runs, his sword magically transforms into a heavy spiked mace, metal shifting and expanding with glowing red energy.

The camera tracks forward with ironclad in a dynamic action shot.

ironclad swings the mace in a brutal sideways BASH, striking entomancer across the torso.

The impact sends entomancer flying backward through the swamp, crashing violently into a large twisted tree. The tree shakes and bark explodes outward on impact.

Water splashes everywhere, insects scatter, fog swirls dramatically.

End with ironclad standing in the swamp with the mace still raised while entomancer is embedded against the tree.

Style: dark fantasy, dramatic lighting, cinematic camera movement, high detail, dynamic motion, game-cinematic style, 4K, realistic physics, slow motion on the impact.

crimson geyser
#

r5 5600x

hallow thistle
low shard
#

you can do either realtime on like discord or just swap the voice out of a pre-recorded audio file

hazy eagle
#

GTX 1660TI 6GB
WINDOWS 10
Wokada Deiteris Fork

while using it, should i turn on HAGS (Hardware accelerated GPU Scheduling) on my windows? or off?

low shard
#

-realtime

patent trellisBOT
# low shard -realtime
🔊 Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

low shard
#

if you need only realtime voice changing, check the pros and cons of both vonovox and wokada tg develop fork and try one of them

hazy eagle
low shard
#

if you need an all in one (realtime voice changing, inference on audio file, rvc training), then try Applio but it's heavier

low shard
#

but I would really suggest switching to vonovox or wokada tg develop first

#

Wokada Deiteris Fork isn't getting updates since a long while

hazy eagle
#

okay thanks

dry thunder
#

what do harmonic strength and voice texture do in vonovox?

#

well thats unfortunate

kind ether
#

hi i got problem

2026-03-14 21:07:56,543 ERROR    [VoiceChangerManager]
Traceback (most recent call last):
  File "voice_changer\VoiceChangerManager.py", line 300, 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 190, in on_request
    result, vol = self.process_audio(audio_in)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\VoiceChangerV2.py", line 133, in process_audio
    audio, vol = self.vcmodel.inference(audio_in)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\RVCr2.py", line 265, in inference
    audio_model = self.pipeline.exec(
                  ^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\pipeline\Pipeline.py", line 238, in exec
    index_audio = self._search_index(index_audio.float(), 8).unsqueeze(0)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\pipeline\Pipeline.py", line 173, in _search_index
    score, ix = self.index.search(audio if self.use_gpu_index else audio.detach().cpu(), k=top_k)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "faiss\contrib\torch_utils.py", line 254, in torch_replacement_search
AssertionError
2026-03-14 21:07:56,606 INFO     [VoiceChangerManager] update configuration serverAudioStated: 0
2026-03-14 21:07:58,751 INFO     [VoiceChangerManager] update configuration modelSlotIndex: 0
2026-03-14 21:07:58,752 INFO     [VoiceChangerManager] Model slot is changed 1 -> 0
2026-03-14 21:07:58,753 INFO     [RVCr2] Initializing...
2026-03-14 21:07:58,753 INFO     [export2onnx] Exporting onnx...
2026-03-14 21:07:58,754 INFO     [export2onnx] Exporting to ONNX on cpu
2026-03-14 21:08:03,337 ERROR    [MMVC_Rest_VoiceChanger] module 'onnxslim' has no attribute 'simplify'
Traceback (most recent call last):
  File "webserver\restapi\MMVC_Rest_VoiceChanger.py", line 98, in post_update_settings
  File "voice_changer\VoiceChangerManager.py", line 251, in update_settings
    self.initialize(val)
  File "voice_changer\VoiceChangerManager.py", line 216, in initialize
    self.vc.set_slot_info(slotInfo)
  File "voice_changer\VoiceChangerV2.py", line 54, in set_slot_info
    self.vcmodel.initialize()
  File "voice_changer\RVC\RVCr2.py", line 67, in initialize
    self.export2onnx()
  File "voice_changer\RVC\RVCr2.py", line 297, in export2onnx
    output_path = export2onnx(modelSlot)
                  ^^^^^^^^^^^^^^^^^^^^^^
  File "voice_changer\RVC\onnx_exporter\export2onnx.py", line 37, in export2onnx
    _export2onnx(modelFile, output_path, metadata)
  File "voice_changer\RVC\onnx_exporter\export2onnx.py", line 146, in _export2onnx
    onnx_model = onnxslim.simplify(onnx.load_model_from_string(io.getvalue()))
                 ^^^^^^^^^^^^^^^^^
AttributeError: module 'onnxslim' has no attribute 'simplify'
#

i am running tg-develop / voice-changer

#

i hope this is the correct channel for this

hallow thistle
kind ether
#

ML*

hallow thistle
#

Send your screenshot to here.

kind ether
#

Logs?

#

I fixed it, I downgraded to the previous version now it works. somehow?

hallow thistle
kind ether
hallow thistle
#

Which Tg Develop's W-Okada version? What I mean by screenshot is an image of your voice changer interface.

kind ether
#

b2397 is the version i had problem with

hallow thistle
# kind ether

Try set sample rate to 48000 Hz, Extra 2.7 s. Also, some people have observed the ONNX model is quite buggy in some Tg Develop's W-Okada versions, DirectML variant especially, only native "pth" files seemingly worked for most.

viral mason
#

Look here and search

lunar trout
#

How do i fix this AttributeError: module 'onnxslim' has no attribute 'simplify'

#

When i first downloaded it it worked fine now it just shows this every time i start

full meteor
#

Guys

#

Huuuuu:3

#

Hiu:3

#

I have a question.

#

How do I create an AI cover model?

#

For the song

#

I don't know who sent this message, but the tutorials written here are very confusing and I understood almost nothing.

#

I looked it up for help. It just says things like "vocal eparation," but I already know that. Other than that, it doesn't explain anything.

kind ether
viral mason
#

Did u rename the file?

#

I don't name my models like that second one

kind ether
viral mason
#

Wdym it's onnx? I have never made an onnx model in my life

#

Did you convert it?

#

The top one is def mine but now I'm confused

low shard
dry thunder
#

im guessing default values are fine for most models?

low shard
#

It was kinda hard to explain them in the docs, but yeah generally it's just best to not mess up with them

dry thunder
low shard
unreal siren
#

-rvc

patent trellisBOT
sharp chasm
#

Yes I need help

clear oasis
#

Can someone tell me how to use it?

#

I'm put the links for the Hug models I want, but it's giving me an error.

#

And testing in a different way, I can get it to work, but it always says that the Index file is missing.

#

I even tried using a ready-made template that someone had already uploaded, adding audio, configuring it, etc., but in the end, when I click "convert," it gives an error "none"

sharp chasm
#

Hi

clear oasis
#

hey

low shard
low shard
patent trellisBOT
# low shard !howtoask
❓ How to Ask for Help
✅ Before You Ask!
  1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
  2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.
📝 How to ask?

Tell your:

  • Full GPU Name: (e.g., NVIDIA RTX 4060 8gb vram desktop)
  • Operating System: (e.g., Windows 11)
  • Detailed Description: What were you trying to do and what went wrong?
  • Tutorial Used: Link to the guide you were following.
  • Screenshot: A picture of the full error message.
🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

  • ANY illegal activities.
  • NSFW/Porn.
    Requests for these topics may be ignored, not helped and result in moderation action.
<:matsuripray:1159685390156967936> Community Expectations
  • Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
  • English Only: Please keep all conversations in English.
  • Don't Ask To Ask.
frank bough
#

One message removed from a suspended account.

kind ether
gloomy apex
urban forum
#

Strange. I am using W-Okada's Fork for AMD GPU's.

#

And the server no longer functions for it.

#

But if I do it via Client it works, but there's no output.

#

<@&1159293204038955078> If any of you wizards or warlock's know what's happening. ❤️

#

CPU: Ryzen 9 7900x
GPU: AMD 9070 XT
RAM: 64 GB DDR5.

#

Whenever I try to run it on the Server, this is the error I get.
The other window just closes.

image.png
#

There also seems to be zero documentation on this error as well. Looked through the chat logs here and nobody seems to have a solution to the xhr poll error.

tame oracle
#

no reason for it

#

it just stopped for everyone basically

urban forum
#

So it's useless now.

tame oracle
urban forum
#

Since I don't know what else is around.

#

That explains why this Discord is deader than a doornail.

tame oracle
#

what were you using

urban forum
#

W-Okada.

tame oracle
#

-# im about to go to bed but curious

urban forum
#

I used the fork.

tame oracle
#

Which fork

urban forum
#

Deit's fork I think?

graceful tinsel
#

@hallow thistle

#

HELP ME TWINNN

#

HOW TO DELETE SHIT IN KAGGLE FOR MORE FILE SPACE

graceful tinsel
#

<@&1159293204038955078>

tight linden
#

hello can anyone help do rvc locally I wanna train but kaggle always glitches for me I have a 10 minute audio file but I wanna speak and then the ai goes over it. I've been trying to do this for months..... lol

graceful tinsel
#

@viscid moss

#

@tame oracle

#

@low shard

#

@cosmic spire

urban forum
#

So it seems AMD users are being hung out to dry with this stuff. Lmao.

cosmic spire
graceful tinsel
hardy yew
cosmic spire
graceful tinsel
urban forum
hardy yew
#

It does so for me

cosmic spire
graceful tinsel
urban forum
hardy yew
#

In the latter case, it will depend on the settings

graceful tinsel
urban forum
hardy yew
#

I believe there is a file persistence setting

urban forum
#

This will be fun to watch.

graceful tinsel
#

where is that

hardy yew
urban forum
hardy yew
graceful tinsel
urban forum
#

The other issue I have with the Deitreis fork or w/e.

#

For some reason now.

#

Client and Server no longer work.

#

Which effectively makes the fork useless.

graceful tinsel
#

and i have persistence variables and files only

hardy yew
#

Well, persistent files means it won't clean them up each time. I don't know the exact rules on how it behaves, I don't run notebooks manually but instead commit them. That means, when you press save version in the top right, you can choose to run it then. In that case, it should start fresh and it will run on its own (you can close the browser). But it's only to be used when your notebook is complete, because it's not editable once you start it, you would have to cancel it and start over with any changes

#

BTW, while tinkering manually, you can just make a cell and run a command to remove the files

graceful tinsel
hardy yew
#

rm -r /kaggle/working/* would remove all contents from that directory

hardy yew
graceful tinsel
hardy yew
#

I'm just getting up, not available

hardy yew
hardy yew
# urban forum Whenever I try to run it on the Server, this is the error I get. The other wind...

Can you add "pause" command as a last line in the app start script? And then make this crash happen again. If the terminal window closes, that means it crashed abruptly. Maybe there's an error log there that you can't see because it closes. If it's auto-paused after exiting, we'd get to see that error.

BTW xhr poll error is just a result of the browser losing communication with the app. Which makes sense because it stops suddenly

chilly furnace
urban forum
hardy yew
#

My suggestion only makes sense for a bat file

urban forum
hardy yew
#

it's an exe, you can instead open a command prompt in that directory and enter a command to run the exe. After the program stops, the logs will still be visible (if there are any)

urban forum
#

How would I go about doing that. I'm not very tech savvy, apologies.

#

If you can ELI5 then I can do it easily.

hardy yew
# urban forum

you can shift+right click the white area below vcclient. Then open a command prompt/terminal/PowerShell, should have at least one of these

urban forum
#

Oh there we go

#

I opened a blue window.

#

That says Powershell.

hardy yew
#

Yeah, now enter MMVC, hit tab, it should write the exe file name

#

Then just hit enter and cause it to crash. See if any info appears in the blue window at the moment it crashes

urban forum
#

Ok. I'll give it a go!

hardy yew
#

Not sure if it actually will with this fork but it it does, that could show us what's potentially going wrong

urban forum
hardy yew
#

😭 it's silent, won't be able to debug it that way then

urban forum
hardy yew
#

I wonder, I've seen people mention using a previous version of tg develop and that it supposedly worked better for AMD. But I'm not sure I remember this correctly, need to search for that discussion

urban forum
#

TG Develop?

#

Do tell.

hardy yew
#

That's the fork you're using I think, no?

#

-rt

patent trellisBOT
# hardy yew -rt
🔊 Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

urban forum
#

Yeah

#

Wait.

#

TG-Fork?

#

I was using the...

#

Checks notes.

#

Deiteris Fork.

hardy yew
#

Ooh. For some reason I was convinced you were using Tg. In that case try it

#

Deiteris is much older IIRC

tight linden
# hardy yew Can you describe your issue precisely?

I wanna take my 10 minute audio file and train it so I can speak and it sounds like the audio file voice I trained. kaggle was always glitching for me though and I gave up on it. also voicemod is a terrible voice changer even if its live

urban forum
hardy yew
#

I mean, tg is not as fresh and maintained as e.g. vonovox, but IIRC it's the best of the AMD-supporting apps

urban forum
#

Vonovox doesn't have AMD support.

#

I'm gonna do a fresh install.

#

I'm on a 9070 XT

#

They all say AMD

hardy yew
#

(although it's worth monitoring the vonovox discord, the dev recently has shown some AMD interest... Obviously wouldn't expect it to be implemented soon, but I think it's on the roadmap)

hardy yew
# urban forum

amd64-dml. I think it's described in the guide I sent above

#

amd64 is CPU architecture, not the target GPU

urban forum
#

Yeah it just breaks on a new install.

hardy yew
#

The same issue, you mean?

urban forum
#

Here. Let me record it.

hardy yew
hardy yew
# tight linden I wanna take my 10 minute audio file and train it so I can speak and it sounds l...

Is your audio data clean and of decent quality? This will heavily influence the chance of getting a good model. Also 10 mins is not a lot, but it can still work.
As a first step i'd recommend reading the guides:
https://docs.aihub.gg/rvc/resources/dataset-isolation/ (general info about preparing dataset and what should be avoided)
https://docs.aihub.gg/rvc/resources/training/ (general info about training)

Dataset & Isolation

Last update: March 11, 2026

Training

Last update: March 13, 2026

#

You have a nice guide how to start the training there

#

With the specifics:

  • sample rate: this'll depend on your input audio. You can check the frequency spectrum with e.g. Spek and double that to reach your minimum target sample rate. If in doubt, 32000 is a nice go-to.
  • embedder: contentvec
  • pitch extraction algorithm: rmvpe
  • batch size: 4
  • epochs: you can choose something like 200, it should be more than enough
  • save every epoch: 10
    Also in advanced check "Save only latest ..." and "Save weight models" or whatever it was called

This'll be a fine starting point, but keep in mind that it doesn't ensure you'll get a model that works and sounds great. This depends on a ton of factors (also massively influenced by your input data)

urban forum
hardy yew
#

Maybe someone with actual AMD experience will comment later

#

My biggest concern is that you said it stopped working randomly and worked before

#

I wonder what could've changed suddenly

#

A windows update? Some driver change?

#

This might be tough to locate

urban forum
#

I'm on an updated AMD driver.

#

So idk.

urban forum
#

Tried everything.

#

In short if you're on AMD. Find another solution.

low shard
# full meteor How do I create an AI cover model?

hello, since i saw a little discussion in #🧬│ai-chat message because there was a confusion between the help and the guides channel, I can help ya right now

You just want to make ai covers like goku sing a song?

It would help me if you tell me your PC Operating System and GPU, so I can tell you your options

#

I just don't want any drama and help everyone

low shard
urban forum
#

The issue is the program just closes upon execution.

mint jasper
#

is there any possible way to remove the delay when talking?

hardy yew
#

Nope, some delay is inevitable

#

it does depend on some stuff, including GPU performance, chunk size etc

#

but it always will be delayed

low shard
low shard
urban forum
#

And I'm guessing it's supposed to close?

low shard
patent trellisBOT
# low shard -realtime
🔊 Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

low shard
urban forum
#

Holy..

low shard
#

this version is what tg develop is based on, it's older and might miss some features, but I just want to check if it works for your hardware or it closes too

urban forum
#

Oh

#

That's the version I am using.

#

That's the one that closes.

low shard
#

this is deiteris, a bit older but just testing if it works for you

urban forum
#

How can I tell which version I am using?

low shard
low shard
low shard
urban forum
#

So which is the one I need then?

low shard
#

we aren't sure, it's just a test

urban forum
low shard
urban forum
#

Oh the green one. I used this for months and then it stopped working.

#

@low shard

image.png
#

When I try to run it.

low shard
urban forum
#

Yeah. One sec.

low shard
#

open again the deiteris one, wait for the error appear if it happens on start up, and show what happens on the CMD window

low shard
# urban forum

can you copy paste the path of where you have stored your folders?

#

perhaps there's a special character in your windows name or the whole path?

urban forum
#

And both of these are in my Downloads folder.

viral mason
#

I saw it wasn't working for you earlier but maybe just reinstalling it fresh could fix any issues

urban forum
#

They were having me try out both versions.

viral mason
low shard
#

then you write MMVCServerSIO.exe

#

then you try to start it

#

this way, it should keep the CMD open

urban forum
low shard
urban forum
fiery raptor
#

is there a way for me to use UVR 5 without removing the back vocals of a song? (only the main)

low shard
urban forum
cinder wyvern
#

is there any free way to remove reverb from vocals without having to pay subscriptions for dumb sites

hardy yew
#

UVR

#

lots of great features and models

hardy yew
#

if your vocals are already separated from music, you can use the dereverb model directly

#

like this

#

also possible to chain it all together, separating vocals from music -> splitting into lead & backing vocals -> dereverbing lead vocals

#

not perfect, but pretty nice, especially for free software

low shard
#

you need a good pc to run such local heavy tools btw

#

don't try it on like super old laptops

hardy yew
#

good point, i never really paid too much attention to how heavy it is on my PC, all i know is that it works fine xD

low shard
#

i feel like most tech people forget that the average person doesn't spend much on a PC unless they play heavy games, so that's why i always suggest staffers/helpers to check the user's hardware first

#

just saying bc it happened i spent an hour trying to help someone just for him to have intel celeron 😭

hardy yew
#

good habit

viral mason
#

what voice changer are you using? @fleet grove , I saw ur message before u deleted it

fleet grove
viral mason
#

personally chunk depends on ur gpu but I use theses settings and it works great

image.png
#

second image is in advanced settings

fleet grove
#

thanks

viral mason
#

you're welcome

cobalt oyster
#

guys I am back after a long time are there new ai innovations like rvc stable diffusion

#

how to make realistic ai images and videos rn what do people use to make ai slop videos

#

vibe code websites

cinder wyvern
#

can i infer a voice onto an audio while simultaneously training another voice on the side

#

or can applio only do one thing at a time

viral mason
fleet grove
#

What's the difference between rvmp and rvmp onnyx?

viral mason
#

regular one is better

hallow thistle
# fleet grove What's the difference between rvmp and rvmp onnyx?

rmvpe_onnx is a variant of rmvpe that's exported in ONNX format to work on non-NVIDIA GPU (Intel Arc, AMD Radeon RX) or CPU scenarios, it can also work on NVIDIA GPU but here's the thing; rmvpe is simply an original one that generally works better on NVIDIA GPU.

fleet grove
urban forum
#

@low shard I GOT IT WORKING!

#

With the help of ChatGPT.

#

I deleted the "stored_settings" folder and it reset everything.

#

Then it began working again.

urban forum
#

Seems to have been an only temporary fix for the client.

#

Not sure about server.

hallow thistle
#

If I ever feel lazy, I might use ChatGPT. nso_ame_shrug

urban forum
#

I'm at a loss. If anyone else is on an AMD GPU who thinks they know, feel free.

urban forum
#

@low shard I think I might've found something.

low shard
low shard
# urban forum

are you perhaps using any audio effects? I would suggest trying to remove any

urban forum
low shard
#

not sure if @hardy yew or @hallow thistle ever saw it before

#

meanwhile i can only say to play out different settings and browsers and gpu drivers versions, i’m not sure

rich saddle
#

what method is the lowest latency and best quality for Realtime Voice Changing? It's been a while since I checked in here, but I have the local okada still.

#

Any one?

hardy yew
#

Vonovox beta if you have an nvidia GPU. Tg develop w-okada fork is also nice (+ has an AMD version if needed)

rich saddle
#

I have a GeForce RTX 4070ti

hardy yew
# urban forum

I wouldn't really bother about the frontend errors that occur when e.g. setting the GPU etc. This looks like bad code that has no impact on functionality

#

And the errors that occur when you start server are due to connection lost because the app crashes

#

So nothing new I'm afraid

#

Still quite puzzling why it does that

#

Can't see a way to do it other than manual build + debugging

#

Because the prebuilt clearly crashes without any error so it's impossible to find the cause from the release

#

Unless someone just knows about particular reasons why it could fail. But there has been no progress since yesterday so it seems to not be a known issue

rich saddle
#

So vonovox would be better than the Tg develop w-okada fork?

low shard
hardy yew
rich saddle
#

I think im still running the old W-okada fork, the original. so anything should be an improvement atp lol?

hardy yew
hardy yew
#

here's a link for latest vonovox beta

#
  • command that needs to be run to fix index (if you want to use index at all)
low shard
hardy yew
#

Yeah, specific to this particular zip

#

Not a general issue

low shard
#

not sure if to add all that info in the docs, since beta versions change a lot and frequently

hardy yew
#

No need to, there will be a polished release soon anyway and this beta won't be useful anymore

urban forum
#

If there's anything new that comes out for AMD, feel free to ping me about it.

#

Since W-Okada doesn't work any more.

low shard
# urban forum

this is mostly just a frontend bug, it happens for me too at start up (but not when i try to connect to the server) but it wouldn't solve the issue since your program actually closes:

  • try again other browsers, and in every browser, switch multiple times between different input and output devices and between client and server audio processing, and try also switching between GPU and CPU
#

did you use this program before and one day it randomly stopped working? if so, do you remember anything you did before it stopped working correctly?

urban forum
#

If I use Client, it will work.

#

But there's no output.

low shard
low shard
#

then try Line 1 as a microphone in discord, and do a microphone test

urban forum
low shard
urban forum
fossil reef
#

I can't create AI Models anymore.

I tried on Colab, Hughing Face and Applio.

low shard
low shard
hot apex
#

i havent used anything related to rvc in a WHILE. is it possible to do TTS with rvc?

#

update: nvm

low shard
#

like Applio does, but it won't be expressive

hot apex
#

but where do i put the audio? im very rusty at this so i dont remember anything

low shard
hot apex
low shard
hot apex
low shard
#

even a different GPU model/branding would cause a totally different guide, AI is more advanced than running a game

hot apex
low shard
hot apex
low shard
# hot apex yea

I would suggest you a Local Program so you don't have to depend on the free tier of sites that are limited, you can use Applio

Applio

Last update: March 7, 2026

#

if you downloaded other software from youtube tutorials, it's better to uninstall it

hot apex
#

how do i download applio? theres no exact link i can find

low shard
hot apex
#

so thats how to set it up, got it

urban forum
#

Does it have the same features as W-Okada?

viral mason
#

applio is mainly to train ai voice models

#

but it can do realtime

urban forum
urban forum
low shard
viral mason
low shard
#

it's harder to install for AMD GPU users on Windows 10/11

urban forum
low shard
#

Nvidia GPUs really have a monopoly 😭, especially about AI

urban forum
#

I have an Nvidia in my other PC.

#

I should look into a 2 PC set up.

viral mason
low shard
urban forum
#

Wait sorry.

#

Let me check something.

#

Oh my 2nd PC has a 1660 Super in it.

#

1660 Super and a Ryzen 7.

#

New PC has a 9070 XT and R9.

low shard
#

if you want, you can still try to use Applio only on your AMD GPU

viral mason
low shard
#

or you could directly just use Wokada Tg Develop or Vonovox on that Nvidia PC

#

the 2 pc benefit is that there will be more load to play games and stream on your main one

urban forum
#

Yeah

#

I need to wait to get paid to get a capture card.

low shard
fossil reef
# low shard are you trying to train RVC v2 Voice Models? what's your pc gpu & os and what yo...

Yes, I'm trying to train RVC Models (there are some of them that I need to create the dataset).

My GPU 0 is Intel(R) Iris(R) Plus Graphics
My GPU 1 is NVIDIA Geforce MX250

Basically, I'm going to make AI Covers (I have Applio Local to make them), but, as I am a gamer and a AI enthusiastic, I will use some of the Models (Spoken Voices) for having fun only for myself.

On that way, for esxample, making people to say like a law while I am playing Pokemon or something like that.

low shard
#

your only way is using one of the cloud versions of Applio in the AI HUB Docs

fossil reef
#

Colab?

low shard
#

the only difference is the free gpu time and the setup between the 3

#

you could check the differences between all 3 services and decide what fits you best

fossil reef
#

I said that I can't on Colab

low shard
#

you got an integrated gpu and an extremely old nvidia gpu

fossil reef
#

I can also train Spoken Models?

On Applio?

Cause I was trying back in 2024 I think, and I was founding like the pitch error.

hardy yew
#

pitch can be adjusted, so not sure what issue you're talking about

#

anyway, you can easily train speech/singing models with applio

#

and if you haven't tried yet, kaggle does the job for me

#

30h weekly for free

viral mason
hardy yew
#

i do like crepes

viral mason
#

that looks gooood

tall haven
#

is it normal for me not to have extra time and crossfade duration im using the bata version of vonovo

hardy yew
#

yeah

#

the sliders were redundant, so got removed

tall haven
#

gotchaa any suggestion on settings even advanced

fossil reef
low shard
low shard
fossil reef
low shard
low shard
eager swallow
#

p

fossil reef
low shard
low shard
eager swallow
#

I'm looking for furina voice

tall haven
low shard
eager swallow
#

okeyy

low shard
eager swallow
#

thankss

low shard
fossil reef
# low shard alr

Yeah, but I'm dowloading the 400 epochs file (PTH and INDEX for testing.

Cause now I'm not going to be at home

brazen lodge
#

Hey! I've been using Deiteris' W Okada Fork for a few days and had no problems at all. I don't know why but since today it isn't working anymore. I've tried switch inputs/outputs, different models, cpu instead of gpu but it doesn't work. Tested if my mic doesn't work or anything, but that's also not the case. I even tried to install it on another windows installation, same thing. Any ideas what it could be?

#

Also tried another microphone..

#

have got the same problem on Tg Develop's W Okada Fork aswell

low shard
brazen lodge
# low shard elaborate: - your pc gpu - your pc os - what you're trying to do - the tutorial ...

GPU: AMD RX 9070XT

OS: Windows 11

What I wanna do: Trying to use the voice changer like I normally do.

Tutorial?: To install it, I used the following tutorial. It worked the last week though without any problems, it's an issue that just started. (https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/)

My Problem: I've been using Deiteris' W Okada Fork for about a week now without any problems, but since today it's not working anymore. I tried different input devices and it seems like it doesn't get any audio from the input since the gpu utilization doesn't change at all when talking. I tried different microphones and a new installation of Windows 11, this didn't solve the problem.

hardy yew
#

sounds familiar

brazen lodge
#

The weirdest thing is that I changed nothing haha

low shard
low shard
brazen lodge
#

And it seems like the audio from the microphone isn't processed

#

(tried different outputs)

low shard
brazen lodge
#

yea sure

agile pilot
#

what program do you guys recommed to use the models on?

hardy yew
#

Another reason to believe it's either a wIndows update issue or (more likely) AMD driver update that screwed it up

#

I'd try downgrading the GPU driver and see if it helps at all

brazen lodge
hardy yew
#

Something changed for sure, considering it worked and suddenly doesn't. The problem is finding what changed exactly

brazen lodge
hardy yew
#

But your issue seems similar to someone else's, this is the second case like that. It might be a different problem, but maybe it's not

brazen lodge
hardy yew
#

However, the other person had the VC crash upon starting conversion

#

Your case is different

#

Because you said there's just no sound but nothing else happens

brazen lodge
hardy yew
#

In tg develop, does the graph in top right appear at all?

#

When the conversion is running

brazen lodge
brazen lodge
#

So the curve goes up at the beginning when u start it, but after it's just a straight line even when talking.

hardy yew
#

alright, this is definitely a different issue than the one i mentioned before

brazen lodge
#

alright

hardy yew
#

in your case everything seems completely fine but there's just no sound for whatever reason

brazen lodge
#

yea

#

it's like it doesn't receive the sound

hardy yew
#

either doesn't receive or doesn't output it i guess

#

you have some VAC, right?

brazen lodge
hardy yew
#

yeah, it should probably be more wavy

brazen lodge
hardy yew
#

I'd try two things for a test:

  • set input to VAC and output to your speakers, then play some audio sample with e.g. audacity (in audacity set output to VAC)
  • set input to your mic and output to VAC and record the VAC in audacity
brazen lodge
#

but I also tried it without the vac

#

same problem

#

so even if my vac would have some problems, this is sadly not the issue

hardy yew
#

yeah i'm kinda thinking out loud as it's a weird issue

brazen lodge
#

okay

#

uh

#

it worked for a short time now

#

dont ask me why

hardy yew
#

😮

brazen lodge
#

tried running my mic through steelseries

#

and then that input to vc

#

worked, then I changed it back directly to my mic and that also worked

#

like nothing happened

#

this makes no sense and is so confusing

patent trellisBOT
brazen lodge
#

So if anyone ever gets the same problem, what I did: Start Steelseries Sonar, use the Steelseries Mic Input as an input for Okada W and change your input back.

turbid adder
brazen lodge
turbid adder
#

yep

#

even reseleted the cpu then back to the gpu

#

cant post screen shots(in this channel) but the setting blow chunk and extra say my gpus name and i have swapped it back and forth

brazen lodge
fossil reef
#

let me know

rare sage
#

Can someone direct me to a video or walkthrough of how to install vonovox please

#

@tame oracle

rare sage
#

I can't get it to work i went through the download im not sure whats wrong

agile kelp
#

how do i use tg develop i got it started on my cloud notebook using kaggle and everything but i dont know how to start the real time conversion

#

There is a button that says start server but i dont think thats how you start the real time conversion

#

cause it says Warning
Select an audio input device. if i press start server

#

even tho i have selected

#

to my microphone

viral mason
agile kelp
#

yes the output its set to my vc cable

#

my input is set to my microphone

#

its checkmarked on client

#

the server checkmark is greyed out

#

saying Server audio not available.

viral mason
agile kelp
viral mason
rare sage
#

Im now using Vonovox and trying to get the hang of it

#

but im struggling a bit with all of the settings its all new to me @viral mason

viral mason
#

do u have the most recent Beta?

#

it's better than the current release

rare sage
#

But im struggling on settings

viral mason
rare sage
#

sometimes it still sounds robotic

#

Im new to all of this i don't really understand settings too well

viral mason
#

there'ss nothing really to it, all u gotta do is change pitch tbh, block size at 0.30 is good

rare sage
#

Yeah i tried .30 but it seems like its working better for me at .50(?)

agile kelp
viral mason
rare sage
#

yes i tried between .30 and .50 , .50 seems to pick up more and cut out some of the robotic ends

viral mason
#

interesting

rare sage
#

but if i go above .50 it gets worse

#

because i tried .70 after and it was much worse than .50 and .30

viral mason
#

weeird

rare sage
#

yeah

#

but it also seems to vary by voice @viral mason

#

what would you say is the best way to get the robotic feel out?

viral mason
#

tbh not sure, it depends on what models you're using

rare sage
#

do you have any that i could try?

viral mason
rare sage
#

u would most likely know the model better than me since you made it

#

so if i experiment with it, if our voices are drastically different and certain block sizes work better for different things

#

maybe it might give us more insight

#

i am having some trouble getting the vonovox 1.7 to link to my discord audio though @viral mason

#

it worked with 1.69 not 1.7

viral mason
rare sage
#

yeah i fixed it

#

just tiny troubleshooting issue

shell birch
#

@hallow thistle

#

Which VCClient file do I download if I'm running intel graphics?

hallow thistle
shell birch
agile kelp
#

switching sio/rest protocols in Tg-Develop does not help

agile kelp
rare sage
#

How difficult is it for me to train my own model if i wanted to train it on like 2 hours worth of talking

#

or more

viral mason
#

-rvc

patent trellisBOT
viral mason
#

use the applio docs and ask here

urban forum
solid mantle
#

I am confused, is the GPT-SoVITS-WebUI the one you can use rvc model to create voice from text?

winter spade
#

Hi everyone, I’m new to AI and recently became curious about Agentic AI and automation systems. My background is in operations, and I’m trying to understand how AI agents are built for real workflows.

Could anyone suggest a good learning path or course to go from beginner to advanced in building AI agents? I want to follow a structured roadmap instead of randomly learning things.

Any guidance would really help

rapid tendon
#

Hey everyone 👋

I'm experimenting with Local AI and trying to run models locally on a GPU with 8GB VRAM.
Right now, when I try to use 4096 context, the model needs around 11GB VRAM, which is too much for my setup.

I'm looking for recommendations for a good uncensored / unrestricted model that can run locally with decent speed on 8GB VRAM.

What I'm trying to achieve:

  • uncensored or low-restriction model
  • works with 8GB VRAM
  • good performance (not too slow)
  • 4k context if possible
  • support for quantization / partial offload / CPU+GPU

What tools or models would you recommend?
llama.cpp, koboldcpp, ollama, LM Studio, or something else?

Thanks 🙏

hallow thistle
#

<@&1159293140440723499> What do y'all think of this one?

hallow thistle
#

Of course, you still cannot directly use GPT-SoVITS models in a dedicated RVC program like Applio RVC, vice versa, unless you were able to integrate GPT-SoVITS features into Applio itself.

low shard
#

!howtoask

patent trellisBOT
# low shard !howtoask
❓ How to Ask for Help
✅ Before You Ask!
  1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
  2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.
📝 How to ask?

Tell your:

  • Full GPU Name: (e.g., NVIDIA RTX 4060 8gb vram desktop)
  • Operating System: (e.g., Windows 11)
  • Detailed Description: What were you trying to do and what went wrong?
  • Tutorial Used: Link to the guide you were following.
  • Screenshot: A picture of the full error message.
🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

  • ANY illegal activities.
  • NSFW/Porn.
    Requests for these topics may be ignored, not helped and result in moderation action.
<:matsuripray:1159685390156967936> Community Expectations
  • Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
  • English Only: Please keep all conversations in English.
  • Don't Ask To Ask.
spare smelt
#

For Applio realtime is there any way to make it save settings like chunk size? I have looked everywhere but cannot find a way to do it

hardy yew
#

I don't think there is, but you probably can just hardcode your settings as a workaround.
The realtime feature in applio isn't widely used as there's much better solutions for that purpose

spare smelt
#

Thank you

viral mason
rare sage
urban crane
#

Hello everyone, can you please help, there is Tsukuyomi-chan a popular project in Japan, I would like to try it for the re-voice of Piper TTS

rare sage
#

does more epoch = better as far as less ai glitches

#

It has more ranges to work with and is less likely to glitch out?

#

If so I could probably just work on my own and train it for hours or days that way it has little issues

low shard
urban crane
low shard
# rare sage does more epoch = better as far as less ai glitches

no, there's no way to check if a model is good unless you listen to samples and try it

longer explaination:
epochs are a unit of measuring the training cycles of the AI model

basically the amount of times the model went over its dataset and learned from it

they don't mean how good is the model, it's just an info provided on how they trained the model by the model maker
More ≠ better
Less ≠ better

There's no way to determinate how good the RVC model is until you try it out or listen to the audio samples if there are

low shard
low shard
urban crane
#

But thanks a lot anyway

rare sage
rare sage
#

The biggest issue i've noticed is that some of the models i find work really well, and others of them sound robotic at times

#

i just want one that doesnt have that robotic feel so i thought i could make my own but its probably not that simple

viral mason
#

It's pretty easy to make your own models if you know what you're doing, just gather audio using YouTube or anything really, use a site like MVSEP or you can use UVR (ultimate vocal remover) locally which idk how to setup since I use web stuff

low shard
#

you could also just try the ones with the most reactions and positive comments

#

but there isn't really a golden rule or parameter to know which model is best unfortunately

viral mason
#

most likely the more recent models of whatever character you're lookin for is the better model

#

not always

#

but most of the time

urban forum
#

Yeah I don't know anything else that's feasible with AMD right now.

#

I tried updating my Python to the latest version.

#

Updated Pip and Pytorch too.

#

Still getting xhr poll error.

hardy yew
#

have you tried downgrading your GPU driver?

urban forum
#

Hold on.

#

Let me actually try that.

#

Wait a minute.

#

You might be onto something.

#

It stopped working after I downgraded my Drivers.

hardy yew
#

we've mentioned tinkering with the driver a couple times before

#

but maybe it got lost between the messages

urban forum
#

The problem is with these new drivers, they cause a crash in my VR.

#

So idk.

hardy yew
#

if i were you, i'd just downgrade even further

urban forum
#

Let me try out the latest update first.

#

Then if need be I'll downgrade.

#

Nope.

#

It just crashes still.

#

Something might just be broken with the code.

#

Which would mean I'd need to build it from scratch.

#

I'm going to downgrade my drivers and see what I can do.

#

Yeah.

#

So I officially throw my hat in with the W-Okada program.

#

It's broken on AMD systems and doesn't seem to work.

#

Client works but doesn't relay audio at all.
Server crashes in it's entirety and doesn't give an error code
the only thing it does is give out the Xhr Error which nobody knows about.

  • I've rolled back drivers.
  • I've used up to date drivers.
  • I've swapped audio devices.
  • I've updated Python, Pytorch and Pip.
  • I've tried resetting/rebooting my PC multiple times.
  • I've run the Deit Fork and the Tg Fork, both crash.
  • I ran Applio and Applio works for real time, but it's bad.
#

The only things I did not try.

  • Using someone else's version of the clients.
  • Build the client from scratch via code.
#

@low shard

hardy yew
#

well, building it manually and debugging is pretty much the only way to precisely locate the issue

#

but I imagine it's probably not an option for someone who's never done that before

#

and TBH "it's broken on AMD systems" is a bit of an overstatement, it doesn't work for you in particular

#

but that's just a comment, not like it matters

urban forum
hardy yew
#

I just like to keep things clear, especially that this is a help channel and some newcomer with AMD GPU could hop in, read the false statement that AMD systems are not supported and come to wrong conclusions

#

Not trying to be rude or nitpick

urban forum
#

If you search "xhr" in the server, you'll see people who came before me, who had my issue, no solution found.

#

I'm not mad at anyone but myself tbh.

#

I just wish I knew what it was, been racking my brain for a solution.
Even asked some coding buddies of mine. They don't know either.

hardy yew
#

If I understand correctly, the xhr poll error is meaningless really, it's a side effect of the app crashing because the browser can't connect to it anymore.
It's not the actual issue to solve

#

The problem is finding out why the app crashes

urban forum
#

Yeah, and that's the issue. Nobody knows why.

hardy yew
#

Sadly, yeah

urban forum
#

So someone would need to essentially re-fork the tool.

#

And fix the issue.

#

There's other solutions out there. Vonorox I think is one.

hardy yew
#

Vonovox is for Nvidia only

urban forum
#

Oh right.

hardy yew
#

Might support AMD some day but definitely not anywhere in the nearest future

urban forum
#

The only thing I could think of is if someone uploads their version of the fork for me to use.
The only thing I can think of is that something happened with the original forks, but that's a far stretch.
Why would both programs old and new fork break respectively when I go to boot them up.

hardy yew
#

they share lots of code so the crash happens somewhere in there, I guess

#

hard to tell

low shard
#

or try with other AMD Drivers Versions, you said Wokada TG Develop Fork used to work before right? so you might need to downgrade more and remember the version you used before

urban forum
#

I'm glad my other PC is Nvidia. I'll buy stuff for it when I get paid next month.

tall bobcat
#

hey guys, is something wrong with wokada? It looks like the voices are lagging or cutting off all the time, this is not an issue I ever had it just happened randomly?

hardy yew
#

Then you can use the Nvidia one for RVC and AMD for everything else

urban forum
#

Wait.

#

That's true.

#

Lmao.

#

I wonder if I can.

hardy yew
#

Just need to make sure there's room for it and that you have all required PSU connectors

#

But usually PSU is bundled with more than enough cables and you should be fine

#

Oh and also obviously the PSU needs to be powerful enough

rare sage
#

like it could work good for the first half of the sentence

#

then near the end it turns into like

#

a slur of words and noises

hardy yew
rare sage
#

So i was looking for ones that don't sound as robot-y

#

or atleast have that issue

hardy yew
#

What index rate are you using?

#

A high value makes issues more likely too

rare sage
#

Im using vonovox index rate 0/ the default because I don't know much about the index rates and don't wish to mess it up

#

Vonovox seems to be working the best for me so far

#

but the robot-yness is an ongoing issue for me

#

I've had multiple friends help me test across calls. They said some of them sound extremely good except for the parts where the words just start to slur and sound automated and i agree

#

so that's the biggest issue im having

hardy yew
tall bobcat
rare sage
hardy yew
#

But also as I said robotic sound is often caused by the model itself too

rare sage
#

Like low quality mic, Compressor,Low pass filter, high pass filter, Eq band 3 and 4. I have access to these but im unsure what they mean

rare sage
#

Its just the fact that it randomly devolves mid sentence at times

hardy yew
rare sage
#

I see

#

My harmonic strength and voice texture are default

#

I know those can make a difference but i heard its easy to mess up also

hardy yew
#

And the remaining things like EQs, compression etc are typical voice processing effects just to tune the sound to your liking. Like adding or reducing some lows or highs etc

rare sage
#

yeah that's definitely true. my only concern is that it will make the sentences where it does sound robotic worse

#

i might want to reduce some highs and keep my voice low

tall bobcat
#

point me out a tutorial at least something

rare sage
tall bobcat
#

i wanna reinstall wokada at least, maybe they had an update?

#

but i cant find any tutorials im using the browser version

#

the newer one

hardy yew
# rare sage My harmonic strength and voice texture are default

Those are in the advanced section, right? In that case they most likely shouldn't be touched and are defaulted to some good values. But can't tell exactly what they do, I don't know. If you want details on it, feel free to ask in Vonovox discord, lots of help available there!

rare sage
#

what do you use instead of vonovox?

hardy yew
rare sage
#

So far with the ones i've used, vonovox picks up and sounds the best (1.7 beta)

hardy yew
#

Vonovox is the best currently in terms of quality

#

Namely the 1.7 beta 11

rare sage
#

yeah thats what im currently using

#

just trying to figure it out

tall bobcat
hardy yew
#

b2332? Is that the mainline w-okada?

#

In that case that's quite old, I'd recommend vonovox or tg develop fork

tall bobcat
hardy yew
#

Vonovox is currently the best quality-wise so I'd recommend that in the first place

tall bobcat
#

im dogshit with programming i dont understand this...

#

downloading now

rare sage
hardy yew
#

Makes it try sound more like the target voice (especially useful if it's a model of some particular character and you want it to sound more legitimate), but quality-wise it's usually worse, prone to glitching and weird issues

#

I think most people keep it low or off

hardy yew
tall bobcat
#

i dont use index anyway

#

dont remember ever touching it

#

so you reckon is the version? its weird to me because up until yday when i last tried it, it was working smooth as always

hardy yew
#

Yeah, in that case you can just extract the zip and use the app right away

hardy yew
#

so more prone to various issues as well as just worse quality-wise

tall bobcat
#

i run ffmpge.exe?

fleet grove
#

Is vovonox better than TG?

hardy yew
hardy yew
#

(nvidia-only FTR)

fleet grove
tall bobcat
# hardy yew just run start.bat

says the publisher could not be verified, then brings me to a cmx.exe that tells me to press any key to continue, then closes

#

oh wait let me extract

hardy yew
#

oh, you need to extract the zip for sure

rapid tendon
#

Hey everyone

I'm experimenting with Local AI and trying to run models locally on a GPU with 8GB VRAM.
Right now, when I try to use 4096 context, the model needs around 11GB VRAM, which is too much for my setup.

I'm looking for recommendations for a good uncensored / unrestricted model that can run locally with decent speed on 8GB VRAM.

What I'm trying to achieve:

uncensored or low-restriction model
works with 8GB VRAM
good performance (not too slow)
4k context if possible
support for quantization / partial offload / CPU+GPU

What tools or models would you recommend?
llama.cpp, koboldcpp, ollama, LM Studio, or something else?

Thanks

viral mason
tall bobcat
#

fuck

tall bobcat
viral mason
#

👍

tall bobcat
#

i mean is it maybe my actual microphone? did it suddenly go low?

#

and it cant detect my voice?

#

but no theres literally no way

fleet grove
#

How come the precompiled vovonox from the guide only has up to 1.7 beta? I thought 1.7 beta 11 was the newest.

viral mason
#

if you want the download to the newest beta tho I have it

tall bobcat
#

there is a big possibility that my mic could have the issue all along, mainly because it is INCREDIBLY too low out of nowhere

fleet grove
#

Thanks though

hardy yew
fossil reef
#

How do I have the priviledge to upload my AI models?

I can't do it on the respected channel from this server.

tall bobcat
#

although for the record i never reduced it

#

but oh well

hardy yew
hardy yew
#

worth keeping in mind whenever something's off about input volume

tall bobcat
hardy yew
#

you mean in vonovox?

#

Defaults, really

#

Might slightly lower the block size if you want to improve delay but that'll affect the quality as well

hardy yew
clear oasis
clear oasis
#

It's only when I try to convert the code that this "Error: None" message instantly appears on the screen.

low shard
#

it's useless to try to fix it

#

it's better you tell your PC GPU and OS and what you're trying to do so we can give alternatives

low shard
remote pecan
#

Hi anyone could help me with this please?

image.png
#

I fixed the error

#

but idk, the voice changer is not opening as app, its only as website

#

and it yapps about something with my graphics card

viral mason
#

do unknow what one u have?

remote pecan
#

wokada

remote pecan
#

cuz like 3 months ago it was app (website app) now I have it as new tab in my browser

viral mason
remote pecan
crisp sable
#

could i create my own model and train it???

viral mason
crisp sable
#

how

viral mason
#

use applio

#

-rvc

patent trellisBOT
crisp sable
#

applio ok

viral mason
#

yep, read through the docs

crisp sable
#

thx

#

do i need to do cloud base training

#

or i cna use my 5090

#

or my i9

viral mason
crisp sable
#

what abt me

viral mason
crisp sable
#

ok ok

viral mason
#

I use cloud since it's easier for me to not take up space

#

idk how to do it locally

rare sage
#

hello

viral mason
#

Hi Marcy

strange folio
#

where do you download vonovox?

#

i need the latest version mine is old

rare sage
#

The best version is the beta version 17

viral mason
strange folio
#

alright imma go to it

#

thank you

viral mason
#

ask for the most recent beta btw since it's better than the current non beta

rare sage
#

most recent beta is version 17? 17.11

#

my file folder is vonovox_beta_17_11 pretty sure thats the most recent

strange folio
#

uh it doesnt seem to be discoverable

#

is there a link?

rare sage
strange folio
#

oh-

rare sage
#

thats the most recent beta

#

If you want their discord

#

it has their discord and patreon linked

strange folio
#

alrighty

#

thank you

rare sage
#

no problem

#

theres also instructions for the download and installation

#

pretty much the same

tall bobcat
#

what do these do?

image.png
rare sage
#

not necessarily a must have/need for a voice to sound good

crisp sable
#

whats feature retriaval rate

remote pecan
#

Im on vonovox gihtub

#

And I cannot find that shi

viral mason
#

lemme send it

remote pecan
#

okii

#

thank you

viral mason
crisp sable
#

gui crashing

#

wont convert

viral mason
#

?

crisp sable
#

rcv gui

#

rvc gui

viral mason
#

are you talking about applio?

crisp sable
#

idk what i have

viral mason
#

oh

crisp sable
#

its RVC GUI

#

i run a btch file and it opens

#

normally converts but now it just crashes

remote pecan
#

I js downloaded some driver

viral mason
remote pecan
viral mason
#

vb cable can cause weird issues on windows and the lite one does not

#

that's it

remote pecan
#

okiiiii

crisp sable
viral mason
#

idk how to use that sorry

#

see if a helper knows more

crisp sable
#

what could i use instead for covers

low shard
remote pecan
low shard
#

Also what's your PC GPU and OS?

low shard
#

Might be easier to understand since it has inages

remote pecan
low shard
#

And fortunately that beta doesn't have any issues from what I have seen recently

viral mason
tall bobcat
#

i just scammed myself, i bought the membership for vonovox and canceled it after so i wont get automatically renewed, turns out that completely revokes my membership

#

can anyone contact the guy in charge of that server and tell them to... do something about it?

#

i got kicked out their server too wtf

viral mason
#

if u got kicked from their server you must have been spamming or saying a bunch of bad things

tall bobcat
#

check general literally

#

i just asked how to unlock premium and then was like ''didnt know i lose membership if i cancel it did i scam myself''

low shard
low shard
tall bobcat
viral mason
tall bobcat
#

yeah its a joke

#

lol

#

im flaming patreon

#

not dr87

#

like i supported the fella, gave him money, ofc im not gonna flame him

#

common sense

low shard
# viral mason

Perhaps a mod kicked him for that? Because automod would have acted instantly

tall bobcat
#

i got kicked like a minute after i cancelled membership

#

maybe it was automatic?

#

i dont think anyone on their right mind would kick over ''wtf is this 2010 ahh subscription trick''

#

like jeez

viral mason
#

no clue

crisp sable
#

RuntimeError: CUDA error: out of memory
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.

tall bobcat
#

i guess... someone pls contact dr87 idek his discord

low shard
low shard
#

!howtoask

patent trellisBOT
# low shard !howtoask
❓ How to Ask for Help
✅ Before You Ask!
  1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
  2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.
📝 How to ask?

Tell your:

  • Full GPU Name: (e.g., NVIDIA RTX 4060 8gb vram desktop)
  • Operating System: (e.g., Windows 11)
  • Detailed Description: What were you trying to do and what went wrong?
  • Tutorial Used: Link to the guide you were following.
  • Screenshot: A picture of the full error message.
🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

  • ANY illegal activities.
  • NSFW/Porn.
    Requests for these topics may be ignored, not helped and result in moderation action.
<:matsuripray:1159685390156967936> Community Expectations
  • Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
  • English Only: Please keep all conversations in English.
  • Don't Ask To Ask.
crisp sable
#

Full GPU Name: GEFORCE RTX 5090
Operating System: Windows 11
Detailed Description: trying to convert vocals with rvc gui with ai model.
Screenshot:

📎 message.txt
tall bobcat
low shard
crisp sable
low shard
#

Uninstall everything from it

crisp sable
#

cooked

#

it was working tho

low shard
#

It's too old

crisp sable
#

i swear it was working

#

on my life

low shard
#

It's better you just stop using that outdated version

#

And use more updated programs that can go faster supporting your GPU

#

That program was made even before your GPU was out, so it's impossible for it to run on GPU for you

crisp sable
#

oh damn it was working tho on gpu mode i converte like 30 songs

#

hella slow tho

low shard
crisp sable
#

ik man

low shard
#

You shouldn't have any waiting issue on such powerful GPU

#

Unfortunately all YouTube tuts are outdated since the AI Cover trends died in summer 2023

crisp sable
#

i downloaded applio just dosnt work

low shard
crisp sable
#

when i click run it stays blank the cmd prompt

#

nvm it works just had to wait

#

thx

low shard
low shard
crisp sable
#

thanks man

low shard
crisp sable
#

ig

#

what is index file