✨│ai-help

[low shard]

Their skill issue atp, but it's also true that maybe the repo could have been named something more specific than "voice-changer"

Moral of the story: people will always use the wrong channels

Btw next time use #🔍│help-w-okada

[junior hawk]

i was told there was a way to get a translator that couldnt play back other people talking in diffrent languages so u could understand

[low shard]

I'm sorry but could you explain what you mean? Like you're using a translator and didn't understand my reply?

[junior hawk]

no no

[rare goblet]

Ayo? @junior hawk level 1 !!!

[junior hawk]

for tts voice wizard i was told there was a built in translator that could pick up peoples in game mic and if they were speaking another language it would translate it for u

[low shard]

Oh I don't really know about that

You can use RVC models in realtime for calls with Wokada, but it won't work as a translator too

[junior hawk]

im talking vr chat

to many ppl speak diffrent languages

[low shard]

Yeah Wokada would work

But not as a translator too unfortunately

I don't think there's a Program that both uses RVC models and is a translator for calls

Afaik

[low shard]

I just did a research, a translator that can help you could be https://github.com/Sharrnah/whispering

It should have an RVC plugin too

I have never used this tho

[quasi dagger]

True, people used to search for models in the get-model-maker-role channel

[low shard]

😭

[thick pine]

Hey guys, I've been trying to test the original repo of RVC-Project and it's web UI with v2 voice models but it always throws an error, the github page says something about additional files that are required for v2 models to work but it's a bit unclear, is that repo not supposed to work with v2 models do I need to switch to something else? Btw I'm running it in docker

[late quartz]

i trained voice but index file not downloaded

its amd problem?

[spice gate]

does anyone know how many epochs are needed for 15 minutes of audio?

[brittle wing]

how i can create the vocal cover?

[restive heath]

using tensorboard to see if its overtraining or not is better idea than fixed epoch

[restive heath]

you need : vocal isolated audio file of a guy u want to be converted
rvc model
and pick 1 of this
applio https://docs.applio.org/applio/getting-started/inference
rvcwebui https://docs.ai-hub.wtf/rvc/local/mainline/#inference-
ilaria rvc zero https://docs.google.com/document/d/1YbXcLFPaGjhOdG5NFkK3QrucCEpHZBwFUxkeMO8aB18/edit?tab=t.0

[restive heath]

can you explain what you did?any log?

[low shard]

What's ur PC GPU

[late quartz]

i run the go web used train mode and i do all the steps but i dont have any index file

[restive heath]

there are 3 train mode
one clicked mode(broken)
train model
train index
u sure u clicked train index?
if yes can i see log?

[late quartz]

i closed log but i will try again train index

[rare goblet]

Ayo? @late quartz level 1 !!!

[late quartz]

yep that worked

[restive heath]

ok

[lusty pilot]

where can I train one?

[coral frigate]

Is there a full tutorial that guides you on making high quality models?

[restive heath]

https://docs.google.com/document/d/1j9J8A8Oop9bMOHmCs3jDXzPujuD6TQ0Q396rJ0MyuIc

[lusty pilot]

Dataset needs to be dry and clean

Also no back voc

[lusty pilot]

What Pre-Train do you use?

[coral frigate]

Literally never touched any of this stuff so I’m figuring it out. When I use okada, my voice isn’t very smooth and sound a bit robotic so thought if I trained my own and merged it with my own voice then maybe it could help. But idk.

[lusty pilot]

Have you tried Google Colaboratory before?

[coral frigate]

I haven’t. Was planning on doing the training locally on my pc

[lusty pilot]

You have to download the pretrain then

You can find one in #1235952130855010365

[coral frigate]

Is that what applio and mangio are?

[lusty pilot]

I suggest KLM 4.2 if you want your model to hit above Head Voice

[coral frigate]

What is a pretrain model?

[rare goblet]

Ayo? @coral frigate level 2 !!!

[lusty pilot]

So you train models without any pretrain? 💀

[coral frigate]

I’ve never trained a model haha. It’s my first time learning how to do it.

[lusty pilot]

Oh

[lusty pilot]

Which software?

[coral frigate]

Mangio I believe but I wasn’t told anything about pretrain models.

[lusty pilot]

I said where can I train one
not how it works

[restive heath]

its pretty much like normal train with masive dataset
if you want to start from scratch remove the default pretrain

[lusty pilot]

no like WHERE can I do it?

on rvcDisconnected?

if yes, how can I download the g and d file?

[low shard]

u need to have a good pc

u cant on colab prolly, bc u would get disconnected after barely 4 hours, and u need alot of hours for training a pretrain

[coral frigate]

@low shard Do you know any tutorials on how to use pretrained models?

[restive heath]

Ah sry i didnt see my message failed to send
I tried once with rvcwebui fine-tuning
Maybe applio works too
G and d are in model's log folder

[low shard]

those are used only when u train a model

are u trying to train a model?

[lusty pilot]

so there's a colab notebook for that?

send me a link

[low shard]

no i just told you u probably can’t for free

[coral frigate]

Yes it’s my first time training a model. I have a clean dataset but I don’t know how pretrains work

[low shard]

its the same way as normal training a model, except longer dataset and lot more of time

u have 4 hours not granted on google colab, so its not good

[low shard]

well pretrains work as a base for your model

What’s ur model language?

[coral frigate]

English

[low shard]

pretrains are mostly used for other languauges as the OG one is trained on english only

[low shard]

if u want u can check https://docs.google.com/document/d/1j9J8A8Oop9bMOHmCs3jDXzPujuD6TQ0Q396rJ0MyuIc

but for english u should be good with alos just the OG pretrain, the default one

[coral frigate]

So is it better to just train a model normally if I want my own high quality model instead of using a pretrain ?

[restive heath]

For me other pretrains i tried gives more distortion than original one

Somehow

[lusty pilot]

can you atleast send me a link to a website where I can train one

[low shard]

if its english u can try just OG one,

If u want u can try Titan or Ov2super but for english, og is good

[low shard]

its the same as training a normal model

same notebook as the one u use for a model

the process is the exact same, just bigger

[lusty pilot]

oh

[coral frigate]

Do you have any advice for how I can make my model as good as it can be?

[low shard]

having clean dataset and using the tensorboard

[coral frigate]

Ty

[foggy cedar]

They say the data set should be 10 hours. I'm wondering if there should be different voices in the 10 hours dataset for finetune

[simple ore]

10 hours is exceesingly large set to train a regular model and probably not enough to train a pretrain using a standard training method

non-standard method may work

[lusty pilot]

Is that the d file and the g file on the model?

[simple ore]

d and g files are weights

you can start with random noise (no pretrain) or with already trained model to train your voice

[lusty pilot]

How to remove the pretrain exactly?
I don't think I can train a model without a pretrain

[simple ore]

remove these values

but all you gonna get is horrible noise

[foggy cedar]

I have a small question, can I do it with apolio? Using the A100 GPU

[rare goblet]

Ayo? @foggy cedar level 4 !!!

[foggy cedar]

I want to prepare a completely Turkish dataset and pretrain with it, maybe this way I think my voice models will be better.

[opal kelp]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[restive heath]

Yeah thats very good gpu
idk how long it takes for u tho
I also rented server with a100 and finetuned it takes few days to do
from scratch even longer maybe 1 week or more i dunno

[foggy cedar]

I will actually train using finetune. is 1 hour enough to start?

[restive heath]

i dont think so
maybe get atleast 10hours

same settings is ok maybe change batch file size from 4 to 8 if u still use 4 batch size and ur gpu have 8gb or more vram

[simple ore]

as long as you have all the required drivers, sure

[simple ore]

it is 32k + bullshit mp3 noise at the top

or whatever that is

[mystic orbit]

where can i find more rvc voice models? i saw a website like a year ago but forgot and this discord doesnt have that many

[restive heath]

weights.gg

[polar quail]

-kaggle

[azure marsh]

📘 Kaggle Notebooks

• Applio Notebook, by Vidal Kaggle
• Applio Notebook, by Shirou Kaggle
• Music Source Separation, by Shirou Kaggle
• UVR5 NO UI, by Eddy Kaggle
• RVC Mainline, by Hina Kaggle
• Original W-Okada's Voice Changer, Kaggle
• Modified W-Okada's Voice Changer, Kaggle
• 🆕 UVR5 UI, by Eddy, ArisDev & Nick088 Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline on Kaggle by Cauthess

Note: Kaggle limits GPU usage to 30 hours per week.

More commands: -colabs, -hf, -local, -rvc, /help

[quartz wyvern]

guys what to do when its not starting: waiting for web server? its counting every ten seconds??

[pastel oak]

Delete the file stored_settings.json

[quartz wyvern]

thanks for your help 🙂

[final shell]

@pastel oakhi help

[pastel oak]

?

ask instead of pinging

[final shell]

ok

the voice changer

isnt picking up the starting of my sentences

probably threshold related issue

how do i fix

[blissful vector]

hello! I'm using RVC v2 (old I know) and I see there's a batch processing section which would be really useful, but I don't know where it outputs to. Can someone help?

[azure marsh]

Hey, onlyin.dreams! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[blissful vector]

!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.

[blissful vector]

hm seems I hit all the marks here

[pastel oak]

if you moved the threshold to the right, move it back to the left

send screenshot of your voice changer aswell + say name of GPU

[steel forge]

What fork are you using?

I suggest upgrading to the latest Applio first. Its super easy to install https://github.com/IAHispano/Applio/releases

also technically the latest version of RVC is V2, but there's different forks that have different optimizations (like the pitch detection)

I'm like 90% sure there's a batch conversion thing. It should output to the outputs folder (the folder structure makes a little more sense on Applio)

[blissful vector]

this one, made a fork like a year ago I think. I remember trying the ilaria one and for some reason I liked this one better.

[blissful vector]

there's a batch conversion thing at the bottom, but idk where the files get stored to https://imgur.com/1sxLXxm

[rare goblet]

Ayo? @blissful vector level 1 !!!

[coral frigate]

Hi I’m trying to make a female voice model and it’s my first time. Am I supposed to use a pretrain model or just train it using my dataset. Does the duration of the set and epoch matter?

[simple ore]

no pretrain: ~50-100hr of audio is required

with pretrain: 10-60m

[final shell]

rtx 4070 super

[final shell]

yea i moved it to left

but im getting wierd background noise

[pastel oak]

enable sup2

i still need screenshot

[final shell]

Ok

[pastel oak]

increase chunk to 120 to give it more time to compute, could be the reason for cutting off in the start potentially

[final shell]

Ok

[pastel oak]

enable passthru (red) and check if your normal voice you hear is cutting off in the beginning too or not

[final shell]

whats that

[pastel oak]

there is a button

passthru

[final shell]

oh passthru ok

yes

its red now

what does it do

[pastel oak]

its your normal voice now

check if beginning gets cut off or not or any other weirdness u notice

[final shell]

yes

[pastel oak]

also i want u to test:

keep speaking without a break, see if any words get cut off
then wait for 10 seconds, and talk again, and if it cuts off again let me know

im thinking your gpu goes idle when you dont talk for a few seconds

[final shell]

now i hear everything constantly

[pastel oak]

then disable passthru again (green) for voice changer and test again

[final shell]

ohh sheeesh

its so much better

W

[pastel oak]

yea your gpu struggled with the 64 chunk then

[final shell]

yea but the perf was green

and it didnt go above 35

[pastel oak]

its still advised to put it above 100 ms to give the gpu more time to compute

imo

[sharp tide]

can anyone assist me in regards to this error with the Voice changer Client -

[checkSamplingRate] Error opening OutputStream: Device unavailable [PaErrorCode -9985]

i have changed the device to multiple things and it wont go awway

[stoic flame]

hey is it possible if u can help me in call

[rare goblet]

Ayo? @stoic flame level 1 !!!

[coral frigate]

whats rhe difference between 32k,40k,48k when it comes to making ai models

[simple ore]

32k training is faster

model trained with 32k produces 32k audio

[coral frigate]

so the higher the better the outcome is?

[stoic flame]

soooo no help lol

[simple ore]

well, there's some difference in training results

[simple ore]

48k is grossly excessive for speaking models

[placid tusk]

what nvidia do i install for low end laptop

[coral frigate]

but is it still a good idea to go for 48 just for the small difference it could make? im making my first voice model and i want to make it as good as i can

[rare goblet]

Ayo? @coral frigate level 3 !!!

[simple ore]

do you actually have 48k audio?

[coral frigate]

i dont think so. no point in going 48k without it im guessing

[simple ore]

generally RVC model has a big issue with clearing the top half of the spectrum

especially if the original audio comes with nothing at all in that half

but then again, you can always train one 32k and one 48k and compare

[brittle wing]

https://colab.research.google.com/github/ShiromiyaG/RVC-AI-Cover-Maker-UI/blob/main/assets/RVCAICoverMakerUI.ipynb found the torture

Slow 🦥

[hearty viper]

so i'm getting 20k ping and i have no reason why making it cut out as well as like 20 second delays

[placid tusk]

im using rvc 2 disconnected, when do i stop training to not run out of the gpu runtime, im training for 500 epochs

also its 18 minutes, is 500 epochs good?

[rare goblet]

Ayo? @placid tusk level 1 !!!

[restive heath]

use tensorboard to see if its overtraining or not maybe better idea than fixed epoch

[placid tusk]

how can i tell its overtrained

i dont know how to use tensorboard, its not activated, can i aactivate while its running

[tawdry vault]

Hi, i made my own voice and am trying to do an ai song thing with it, how do i convert the mp3 file into a voice model so it registers with rvc?

[restive heath]

i wouldnt recommend using mp3 i like to go for flac or wav that is lossless quality for datasets for making rvc model but
What's your gpu?

[tawdry vault]

Rx 7900 xtx

Btw ive never done this so

[restive heath]

if you want to train locally you can try
using applio
first learn to isolate and clean audio
https://docs.google.com/document/d/17fjNvJzj8ZGSer7c7OFe_CNfUKbAxEh_OBv94ZdRG5c/edit?tab=t.0
or skip to
https://docs.applio.org/applio/beginners/first-model
this should cover everything you need to know

[tawdry vault]

Im too dumb for this

[rare goblet]

Ayo? @tawdry vault level 1 !!!

[restive heath]

i sent simple way to do in ur dm

[coral frigate]

I’m using mangio to make a voice model. If I want to use a pretrained model then do I just need to add it to the base model G or D path alongside my dataset?

[restive heath]

Put them into the pretrained_v2 folder in your RVC.
make sure to put version v2 and set the G & D pth of the pretrain you downloaded and want to use.

[coral frigate]

Could you help me figure out how many epochs I would need if I’m training a 15 minute audio with a pretrain. It’s my first time doing it

[azure marsh]

Hey, KIWI IS DEAD! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[restive heath]

i cant tell extract epoch
use tensorboard to see if its overtraining or not

[coral frigate]

Does the number of cpu processes effect the voice quality at all when training a model or is it just how fast the model will be trained

[restive heath]

it just effect speed
while training it uses gpu tho

[coral frigate]

Ow ok thank you

[knotty moth]

that option is only applied on feature extraction stage and there's no point of going more than 2 regardless how many cpu cores (also avoids freezing/crashing the system)

[shadow mirage]

hi, have problem, all voices have a robot voice, how fix?

[pastel oak]

If youre using realtime voice changer, bad model for realtime. Look for streamers, egirl, mommy voices

[tranquil void]

hi there can i know what does this mean? and does this effect anything when u turn it on and off

oh wait i dont have pic perm

its basically chinese wordWASABIchineseword right above input device

[pastel oak]

What program, are you using realtime-gui

[tranquil void]

ya

[pastel oak]

Its about exclusive wasapi mode which can achieve lower delay as long as you select windows wasapi for both input and output. But it can crash the program, theres an extra step needed

https://rentry.co/lessdelaywasapi

[tranquil void]

does it work for asio? cuz im using it rn

[rare goblet]

Ayo? @tranquil void level 1 !!!

[tranquil void]

oh i can post pic now

@pastel oak

[pastel oak]

Asio already has lowest delay

[soft mirage]

can i get help in vc by someone?

[restive heath]

what happened?

[soft mirage]

i dont know how to download the voice changer 😭

[rare goblet]

Ayo? @soft mirage level 1 !!!

[soft mirage]

@restive heath

[restive heath]

whats ur gpu?

[soft mirage]

rtx 4060

@restive heath

[restive heath]

https://huggingface.co/Shadicti/deiteris-Fork/blob/main/voice-changer-windows-nvidia-b2309.zip

download this

[soft mirage]

okay thanks man

[restive heath]

here is guide https://rentry.co/ForkVoiceChangerGuide

[soft mirage]

alright appreciate it

[soft mirage]

hey i still cant get it right i keep doing something wrong u got time to help by any chance?

[restive heath]

what's wrong?

[soft mirage]

i cant seem to understand these guides although i have downloaded it the guide doesnt seem to work with my understanding

[restive heath]

can you explain where you stuck?

ok so
you extract the folder
run this .exe
set F0 to rvmpe
set extra to 2.7s
set chunk to 100ms
start and wait till this

window popup

if it not stable at green color increase ur chunk

[restive heath]

or where do you stuck at?

[soft mirage]

can u vc ? u can be muted its ok

[restive heath]

ok

[soft mirage]

i cant screenshare

in this chnanel

[restive heath]

shoot me dm

[soft mirage]

ok

[distant turtle]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[opaque loom]

-audio

[azure marsh]

📚 Audio Guides & Tools

• Creating Datasets for RVC using iZotope RX11, by Cauthess
• Gathering and Isolating Audio, by SCRFilms ❄
• Instrumental and vocal & stems separation & mastering guide, by deton24
• Vocal Mixing Tutorial, by Roomie
• https://mvsep.com/

More commands: -colab, -uvr, -karafan, -overtrain, /help

[honest ibex]

guys when i run rvc-gui.bat nothing happens

[spring orchid]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[low shard]

rvc gui is outdated

do NOT follow yt tuts

What's ur pc gpu and what are you looking to do?

[glacial kiln]

please how can i create an ai voice model myself?

[rare goblet]

Ayo? @glacial kiln level 1 !!!

[turbid gyro]

what should i use

[low shard]

#✨│ai-help message answer my questions

[low shard]

what's ur pc gpu?

[glacial kiln]

8gb NVIDIA RTX 4050

[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

ofc chek the docs in general too https://docs.ai-hub.wtf

[glacial kiln]

Thanks.

[low shard]

yw

[frank tinsel]

Hello, which version of W-okada is the most stable? (windows)

[coral frigate]

No for model training what’s the easiest way to download and launch tensorboard? I can’t find any easy tutorials

[low shard]

use the wokada fork that has more performance, what's ur pc gpu?

[low shard]

are you doing it locally?

[coral frigate]

Yes I’m using mangio locally

For training

Locally

[low shard]

mangio is kind of outdated

what's ur pc gpu?

[coral frigate]

4090

[low shard]

u could try https://docs.ai-hub.wtf/rvc/resources/epochs-tensorboard/#tensorboard

but i'd suggest u switching to applio or mainline

Never follow yt tuts btw

[coral frigate]

Thank you. Why is applio and mainline better? Do they help with quality of models?

[low shard]

the quality is the same to all RVCs

but mangio is outdated since a year

applio has the easiest UI btw, and is slightly faster

[coral frigate]

Ow ok. Thank you

[idle osprey]

rudylis5 if i ever catch you...

[coral frigate]

Should I still try using a large dataset even if I’m using a pretrained model for best results?

[rare goblet]

Ayo? @coral frigate level 4 !!!

[distant hamlet]

og pretrain>>>

[coral frigate]

what does that mean?

[distant hamlet]

og pretrain is always the cleanest. just focus on a quality dataset - doesn't even have to be that large tbh

[brittle wing]

How much batch size for 5-10 data

[restive heath]

5-10 min? 4 batch size

[brittle wing]

Cause that person has only one song

Mitsaytold me to stick to 8

[restive heath]

that works

[brittle wing]

8 works for small data?

[restive heath]

yeah but i like to go for 4 batch size its better for < 10 min dataset for me

[knotty moth]

may be fine for 5 mins/more

[sullen scaffold]

Hello, Im trying to install the software from gifhub but it's not working, I executed the .bat file and it won't work, it just shows errors, and the program won't start

Can anyone help me?

[azure marsh]

Hey, Belmont! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[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.

[rare goblet]

Ayo? @sullen scaffold level 1 !!!

[sullen scaffold]

@long forge may you help me?

[long forge]

What software are you referring to?

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[sullen scaffold]

RVC,

I try executing GUI but it shows errors

I extracted it from the ZIP, and I couldn't open it, so I sent it to Desktop to see if it would work, and it didn't, it kept showing those tabs with errors

[long forge]

https://github.com/Tiger14n/RVC-GUI

Is this it?

[sullen scaffold]

Yes

Sorry for the late responses, it's pretty late here

[long forge]

I think you should look at this page and try again.

https://github.com/Tiger14n/RVC-GUI/blob/main/README.md

[simple ore]

also try extracting it to a folder that is not stolen by OneDrive

[jaunty shale]

So I have a little problem when it comes to training a voicemodel and I cannot figure out why.
I grabbed original voicelines ripped from a game and put them all in one. Which is 16 minutes long and 20 seconds.
I checked sample rate which is 48k and I put those in audacity once I was done merging the voicelines.
These are the settings. (saved in FLAC)

I use RVC Disconnected where I load it in there as usual. put sample rate to 48k and set pretrain to original and batch size to 8.

It does a good job training a model. The problem is, the voice itself is not very crispy and clean when it goes high, especially with singing.
I don't know if it's because of pretrain model or something else. I can provide an example if anyone's interested to help.

I use applio for conversion

[simple ore]

send dataset

[jaunty shale]

at the end when he goes higher, it's not clean enough, like you can hear a slight artifact when you pay attention.
this wasn't the first time this happened, I tried mine roughly half a month ago and it was still doing that.

[simple ore]

it is robot voice coming out as the source does nothave such range

[jaunty shale]

it's not only that, like I said, I tried it with my voice with enough range and it was still doing it

it wasn't doing it before.

[simple ore]

lemme try something, it gonna take a bit

[jaunty shale]

ok

[low shard]

RVC GUI is very outdated

don't use that nor follow yt tuts

what's ur pc gpu?

[frosty python]

Hi guys, i am using so-vits 4.2.26 and got permission 13 error, anyone familiar with the issue?

[rare goblet]

Ayo? @frosty python level 1 !!!

[short heath]

SOFT GIRL UWU LINK NOT WORKING

PLEASE HELP

[rare goblet]

Ayo? @short heath level 1 !!!

[low shard]

so vits svc is old

RVC is better

what's ur pc gpu

[frosty python]

i used 4090 laptop, im trying Applio now

[simple ore]

well, I dont know the name of this type of voice

[jaunty shale]

It's Soldier. 76 from Overwatch

[simple ore]

no, i mean this kind of pecular characteristic

[jaunty shale]

growly old voice I think-

if that's what u mean

either way, it does it even when I have a good dataset for high and low pitch.
It was happening before I discovered that I should change sample rate to match it with the dataset

[simple ore]

so the singing comes out quite robotic

this growly effect makes it much worse

but normal speech is okayish

[jaunty shale]

well he has enough scream in it.

when I looked at others, their voice models are fine no matter if it's growly or not

like the model itself is fine, it's just the consistency of the high note is little... distorted

I don't know if applio does that or if it's because of the pretrain type

[simple ore]

I did not use an index in my inference

[jaunty shale]

neither did I on the result audio

[simple ore]

the range is decent

but this growling is too much

[jaunty shale]

I remember someone making a voice for ramattra and the growly is almost same

the high notes are just fine

difference is the sample rate

it happened with my voice too

[jaunty shale]

I deleted it because I was frustrated that time

[simple ore]

try preprocessing without 'process effects', the filter maybe messing up with it

[jaunty shale]

that's on applio I believe?

[simple ore]

yeah

[knotty moth]

-gui

[azure marsh]

https://cdn.discordapp.com/attachments/1122285248844144733/1203460490475343953/caption.gif?ex=65d12cec&is=65beb7ec&hm=bd2fb8d010006dd7c6e3c1c67d3ae846fd1478e1a3124c544c31b43086fe54aa&

[knotty moth]

nah that application is too old and doesn't have rmvpe

use Applio or mainline RVC1006 for running locally

[frosty python]

uhh, i got this error NoneType' object has no attribute 'pipeline running applio locally on web ui

[simple ore]

select the voice model bud

[frosty python]

oh damn, there was a typo

so, my bad but i get these instead, self.tgt_sr = self.cpt["config"][-1], wrong model type?

[simple ore]

something fucky with the model

[jaunty shale]

I used 3.2.1 applio instead that time, but I can try 3.2.6.

[simple ore]

the last value in the config dictionary is the sample rate

[jaunty shale]

reason why I used 3.2.1 is because split audio caused quiet parts (mostly at the start) completely cut off and some vocal singing wasn't in sync.

[simple ore]

use your own splitter then.. audacity or something else

[jaunty shale]

I meant this

splitting manually is time consuming and it had good results back then either way

[brave swallow]

,

[simple ore]

i can give you a splitter

[frosty python]

i see, so i need to try another model to see if it works

[jaunty shale]

?

[simple ore]

it breaks the file into 3 sec chunks with some small overlap

edit the path where to save the result (at the end), then run from applio folder using env\python split_audio.py

[jaunty shale]

oh wait, where do I put that file

[simple ore]

into applio folder

[jaunty shale]

where .bat files are

[inner sequoia]

Hi everyone,
We’re thrilled to announce that an amazing AI-powered app with two powerful tools "Nacho" and "Musico" to help you master in dance and singing , is launching soon!
"Nacho Tool": Learn dance with step-by-step guidance! The AI shows your mistakes frame by frame, helping you perfect your moves.
"Musico Tool": Improve your singing with real-time feedback! It uses graphs to show where you’re off pitch and helps you stay on tune.

If you're interested, fill out this form to share your feedback and be the first to know when we go live.

Stay tuned, and let’s dance and sing together soon! 🎵💃https://forms.gle/5j9vkRW4yq6d8JUdA

[simple ore]

dont use so-vits models 🙂

[simple ore]

yes

[jaunty shale]

alright-

I put it there and ran applio

[simple ore]

no

edit the file, change the path to your .flac and where to save the chunks

[cobalt pulsar]

Hey im having an issue with virtual audio, its my first time using it. I downloaded the one that was asked for me to download on the website. But im not sure if virtual audio is picking my sound

[simple ore]

run it from applio folder from command line using env\python split_audio.py

[cobalt pulsar]

after downloading was i meant to set it as default or something? or just leave it as be

[jaunty shale]

?

[simple ore]

The file I gave you have nothing to do with applio, other than it uses the libraries installed with it

[ruby kiln]

where is shlattM

[cobalt pulsar]

the highlighted part i was confused about so didnt do.

[rare goblet]

Ayo? @cobalt pulsar level 2 !!!

[cobalt pulsar]

i did click yes tho

[simple ore]

👉 https://discord.com/channels/1159260121998827560/1159290161683767298

[jaunty shale]

Oh! I know what u mean

[jaunty shale]

oh nvm

I got it

forgot to add another folder

[simple ore]

change it to an existing path from which you gonna load those files into Applio

once it is split, use preprocess, uncheck split audio and uncheck process effects, then load it

[simple ore]

yeah... the filter fucked it all up

much better without

[jaunty shale]

for that I have to do it in applio?

[jaunty shale]

ok done

oh wait

[opal kelp]

I'm making a Barney Guttman model where I'm going to use some lines and also some songs where he sings, I wanted to know if this could cause a problem or not using a speech dataset and a singing dataset.

[simple ore]

using filtered set

using unfiltered

[jaunty shale]

oh now I see the difference

wait, can I use those chunks (dataset) in rvc disconnected?

[simple ore]

maybe?

[rare goblet]

Ayo? @simple ore level 38 !!!

[jaunty shale]

I don't have the best GPU

so I'll try

[simple ore]

you can zip the chunks and load them to rvc disconnected

but there are no preprocess settings, it may still fuck them up

with slicing and filering

[jaunty shale]

I actually took the ones already made from preprocess

from here

[cobalt roost]

Which repo should I use for realtime voice change?

[restive heath]

whats ur gpu?

[cobalt roost]

3070

[restive heath]

I like fork you can try https://rentry.co/ForkVoiceChangerGuide

[cobalt roost]

Thanks!

[rare goblet]

Ayo? @cobalt roost level 1 !!!

[royal vapor]

How can I find the older models when selecting a category? I get only recent and a couple more

[coral frigate]

Hi I just trained my own model but it still has room to train I think. How do I continue training it after it’s done training ?

[restive heath]

put in same model name same settings increase ur epoch and it will resume

[coral frigate]

Do I need to add the pretrain thing from the specific log too? The g and d file

[simple ore]

resume starts from the model's d/g

pretrain is no longer needed

[coral frigate]

So I don’t need to add the d and g file? It will just continue from where it ended if I add the settings? Would it also be ok to increase the batch size so it continues faster?

[restive heath]

it should be fine

[brittle wing]

how i can create mine ai voice model?

Program askin' for zip archive of model, but idk what to put in: just mp3 file with voice, or smth else?

[brittle wing]

https://github.com/Tiger14n/RVC-GUI this one

[coral frigate]

does this mean that my model is overtrained? the graph is looking weird

[simple ore]

you have restarted from way older checkpoint

so the chart is fucked now

[ripe oasis]

be on a model with a voice

[coral frigate]

It was fine until it reached 120k. Would It still be ok to use the latest trained model when finished?

[simple ore]

did you delete some latest D/G files?

[coral frigate]

i didnt delete anything. i just continued it but changed the epoch to 1000 and the batch size from 4 to 12

and it continued from the epoch it finished at so i assumed it was fine

[ripe oasis]

what time is it

[coral frigate]

what do you mean?

[ripe oasis]

people hate me 😢😢😢

[coral frigate]

i love you

[simple ore]

okay, changing the batch size did it

so stop this nonsense

[ripe oasis]

why people hate me

[brittle wing]

How i can get ZIP archive with .index and .pth file from my mp3 (voice)

?

[ripe oasis]

Why do you love me?

I just hate others 😢😢

[brazen merlin]

Hello could someone work me through the steps to make ai voices for Mac

Trying to make voices like this

Possibly singing too

Like this

[brittle wing]

Cmoon, how i can train my own RVC model with only vocal audio file? 😭

[rare goblet]

Ayo? @brittle wing level 1 !!!

[low shard]

what's ur pc gpu?

[brittle wing]

If I understood the question correctly, then 16 GB

[low shard]

i mean the GPU name, and also the gpu memory

[brittle wing]

Uuuhhh

second

[low shard]

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

and well in the same GPU tab u can find the gpu memory too

u might have smt like GPU 0 and GPU 1

[brittle wing]

lemme find it, im on russian lang, so there is no "GPI"

u*

yo

[brittle wing]

NVIDIA GeForce GTX 850M

inter(R) HD Graphics 4600

intel*

[low shard]

that's bad unfortunately

[brittle wing]

💀

[low shard]

your gpu is too weak to train (make) models locally (on ur pc)

you can use cloud (remote good pc) tho

[brittle wing]

well, lemme do smth elsesecond

second

[low shard]

?

I can give you the Cloud Ways to run RVC

[brittle wing]

wait a few second, i'll ask friend's gpu

he also got installed rvc rn

maybe his gpu can hold this

[low shard]

As you dont got a good PC, its better you use cloud (remote good pc) for training an RVC Voice Model:

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

Google Colab = easier, but low gpu time, so risk of getting disconnected
Kaggle = Harder, but more gpu time

[brittle wing]

it

[low shard]

Those are the cloud ways

let's wait for what ur friend says

[brittle wing]

kk

@low shard

[low shard]

that's a CPU, not GPU

[brittle wing]

uh

well

[rare goblet]

Ayo? @brittle wing level 2 !!!

[low shard]

u could try asking your friend what's his GPU instead

or use the Cloud ways

[nocturne lintel]

Pipeline is not initialized. What is this?

[quick sun]

I have model Files how do I use them?

[low shard]

for what? AI Covers or realtime voice changer for calls?

and what's ur pc gpu?

[quick sun]

4080 super

[quick sun]

covers

[low shard]

your installation seems corrupted.
try to delete pretrain and model_dir folders and relaunch.

and next time use #🔍│help-w-okada

[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

[true yoke]

I was just about to ask that. Thanks

[rare goblet]

Ayo? @true yoke level 1 !!!

[low shard]

huh?

[true yoke]

I was gonna ask for some local tools since rvc gui is outdated

[low shard]

oh lol
btw the RVCs that I sent are only for training (make models) and inference (use models) on pre-recorded audios, not realtime
(just saying in case u needed realtime)

could i ask what's ur pc gpu?

[true yoke]

AMD. Probably not supported. I use my cpu

Rx 6600 to be specific tho.

[low shard]

AMD is supported

let me share u the AMD version

[true yoke]

Wait fr? Thanks in advance

[low shard]

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

this is for windows

[true yoke]

Thanks

[low shard]

you're welcome

[polar tendon]

Did pytorch break on cloud servers or is that just me?

Because it can't read something

[low shard]

yea your dedicated GPU memory is actually 6gb

[brittle wing]

is that good or bad

[low shard]

usually it's suggested to have 8 or more dedicated gpu memory

[terse shard]

Hey what's the latest version? And is it worth updating to?

[low shard]

@brittle wing it's still doable locally (on ur pc)

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

im not really sure for that tho

[brittle wing]

mhuh

[low shard]

u might need to run on lower batch sizes

[low shard]

latest version of what

[low shard]

what are you using?

[brittle wing]

isnt there like a program online to do this

[polar tendon]

Anything that uses RVC at all

Same error

[low shard]

Yea there are cloud (remote good pc) ways, give me 1 sec im gonna send u them

You can use cloud (remote good pc) for training an RVC Voice Model:

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

[low shard]

i think you are using an outdated yt tut or some old huggingface spaces

you are looking to make ai covers right?

[polar tendon]

AICoverGen, rvc_zero, Applio i think, anything that uses rvc gives the same error when inferencing

[low shard]

yea i think ur using older versions, for example i remember the aicovergen huggingface space is broken

[polar tendon]

I tried google colab too

[low shard]

btw, could i ask you what's ur pc gpu first? as you might be able to do it locally instead of cloud

[brittle wing]

i wanna make a model of the consul from the half life 2 beta

i got the samples but idk what to do after that

[polar tendon]

I don't have a PC with a GPU (only one with no CPU GPU driver)

[terse shard]

RVC

[low shard]

oh nvm then

You can use cloud Cloud (remote good pc, easier and faster than ur PC but it's limited usage):

• Ilaria RVC Zero: fastest and simplest that you can get for free
• Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
• Applio Colab: max 4 hours, not granted, of GPU

[low shard]

like locally to install on your pc?

Well either mainline or Applio

those are the updated versios

[polar tendon]

I'ma spend 7 minutes installing it or use a version that allows inference via CPU

By installing i mean on colab

[low shard]

Use the ones i told you

[terse shard]

Yes currently using this one but I believe it's outdated

[low shard]

You can use RVC inference on CPU if u got atleast 8gb of ram, but it will be slow asf, it's way better if u just use the cloud ways i told u

[low shard]

that's not RVC

that's Wokada that uses RVC model in realtime for calls

its 2 different type of programs

[terse shard]

Ah heck

[low shard]

Well, the most updated version would be the Wokada fork

-rt

[azure marsh]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[low shard]

1st link

it got also better performance

for wokada next time use #🔍│help-w-okada

[terse shard]

Thank you.

[low shard]

You're welcome

[finite barn]

I'm at my wits end with colab and I don't know how to identify the issue I'm getting.
I was able to run the vanilla RVC web UI just fine a few months ago, but anytime I try to run it now, any attempt at inferencing causes an error bubble to appear. I tried running the manual inferencing script, but it complains about missing the "VC"/vc_infer_pipeline module, and I'm not sure what the equivalent to that is in the webui code to install. I had images but I just left the RVC server because no one there was really able to help me despite that being the server their repo linked to

I've tried running in verbose mode with python and there's no error output, so I suspect it's a gradio ui issue, but I've never worked with gradio so I don't know how to fix that

[low shard]

You can run RVC on cloud (remote good pc):

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

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

Google Colab = Easier but risk of getting disconnected
Kaggle = Harder but way more gpu time

3. Be sure to know about the tensorboard

If you are looking for the easiest way and for free, is using https://weights.gg which ofc uses RVC

[low shard]

what google colab link are u using

[finite barn]

This is an oddly common question. I'm using my own version of the notebook

[low shard]

I mean you can just use the notebooks made already by other people

there isn't really a need of coding one yourself tbh

[finite barn]

I didn't "code it" myself. I modified the vanilla RVC web ui notebook

[rare goblet]

Ayo? @finite barn level 1 !!!

[finite barn]

and this setup worked just fine a few months ago

[brazen merlin]

Can someone help me with the steps for mac?

[azure marsh]

Hey, drez! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[brazen merlin]

!howtoask mac rvc help

[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.

[low shard]

What are you looking to do?

[low shard]

Ah

[brazen merlin]

This exactly.

but want to do some voices in differnet dubs

[low shard]

Btw what was the modification you did?

[finite barn]

https://colab.research.google.com/github/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/Retrieval_based_Voice_Conversion_WebUI.ipynb This is the notebook I referenced when I made my own and this is the notebook I was using a few months ago

[brazen merlin]

ill probably have to get the dub voice and then vocal isolate it.

[brazen merlin]

idk what you call this rvc??

text to speech??

[low shard]

You'd need to:

1. Isolate the instrumentals
2. Run RVC either locally or cloud
3. Record yourself saying the things
4. Convert your voice to the RVC model one

This is the way you can do it for RVC

you can also try Applio's TTS tho

it doesn't have really emotions tho

[brazen merlin]

Oh I don't wanna record myself lol

unless that would make better results.

[low shard]

if u want u can try Applio's TTS which has good quality and is multilingual, but its emotionless

[brazen merlin]

Idk if recording myself would work if I'm a guy so

[brazen merlin]

where do i find applio tts?

[low shard]

as RVC is originally made for Speech To Speech, not Text To Speech

So what Applio does is making a TTS audio file with Edge TTS, then convert it with RVC

[brazen merlin]

So what would be the best based on the vid I showe?

[low shard]

https://docs.applio.org/applio/getting-started/installation

[brazen merlin]

I'm on mac tho

[low shard]

you could try with their TTS,
if it doesn't give good enough results you can also voice act it then convert your voice ofc

[low shard]

Ah no one uses the notebook which you are based on tho

[finite barn]

There's no console errors and gradio doesn't have an error log to read

[brazen merlin]

i follow the steps but it says this when i paste code in terminal. chmod: run-install.sh: No such file or directory

zsh: no such file or directory: ./run-install.sh

[brazen merlin]

How do i do that

do i drag the whole thing into terminal??

zsh: permission denied: /Users/dre/Desktop/Applio-3.2.7

[finite barn]

It doesn't really matter, does it? It worked just fine before and now it doesn't

[brazen merlin]

I follow steps but it is not making sense

[low shard]

no, you just open cmd cd Applioand then run the command

Or if you find the installation too hard

You can use it on Cloud (remote good pc, easier and faster than ur PC but it's limited):

• Ilaria RVC Zero: fastest and simplest that you can get for free
• Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
• Applio Colab: max 4 hours, not granted, of GPU

[brazen merlin]

what does cmd cd mean

[low shard]

do you have colab pro? because that one doesn't have encryption, meaning that you can get disconnected as rvc is banned on the free colab tier technically

[finite barn]

Yes, I have pro

I'm aware that it's not permitted on free tier. All of these scripts go to great lengths to obfuscate urls and the like. Very irritating, if I'm being honest

[brazen merlin]

stil says this. zsh: no such file or directory: ./run-install.sh

[low shard]

I'm not on mac but you should be able to by right clicking on the applio folder then Services → New Terminal at Folder

[low shard]

What modification are you trying to run btw?

[finite barn]

No modification. I'm just trying to get the UI to work properly

[low shard]

I mean the modifcations you said you made to the original notebook

[brazen merlin]

Installing Command Line Tools for Xcode

says this. does that mean I wait??

or?

it said this at the end. ./run-install.sh: line 161: brew: command not found

then it said this. ./run-applio.sh: line 3: .venv/bin/activate: No such file or directory

[low shard]

could you show me a screenshot?

[brazen merlin]

yes

not sure what to do next??

[rare goblet]

Ayo? @brazen merlin level 5 !!!

[brazen merlin]

arclame why you yelling at me

[low shard]

nvm it's better u either wait for someone that knows better about Mac (like @proper shale ) or Use the cloud ways i told u

Sorry but idrk about mac

[low shard]

it's just a bot telling you that you levelled up in the server

[brazen merlin]

whats the best cloud thing to use

[finite barn]

I cleaned it up a bit and reordered the code blocks, but that's it really. The process is still the same as it was for the original notebook, except I updated the file paths and removed the manual Aria download section because it's supposed to do !python3 /content/Retrieval-based-Voice-Conversion-WebUI/tools/download_models.py instead

[rare goblet]

Ayo? @finite barn level 2 !!!

[low shard]

#✨│ai-help message all of those have the same quality, the easiest is weights.gg

weights.gg has also a TTS itself iirc

[brazen merlin]

ok. can i make my own voice for it too.

[low shard]

You can on weights.gg

there are also other cloud ways which are harder tho

want me to tell you anyways or nah

[brazen merlin]

idk i just want the voice to be accurate

[brazen merlin]

what about ai images for mac?? is that easier? i want a free one.

[low shard]

Oh, sorry but I can't really help you except suggesting to use Applio Colab or Ilaria RVC Zero for making AI Covers
the notebook you're based on isn't updated since 10 months so

[finite barn]

You should take a look at the one I've actually been using rather than the one I based it on. I did upload my edited one

[low shard]

Mac isn't really the best for AI tbh

the best Open Source AI model for free would be Flux.1-dev, you could try either try running it locally with https://github.com/divamgupta/diffusionbee-stable-diffusion-ui

Or using https://weights.gg which is still the easiest way

[low shard]

I don't pay for colab so I can't really test it even

Tbh I'd just go with Applio Colab, you don't even have to worry about getting disconnected + it's faster and has a simpler ui

[brazen merlin]

oof ok.

[low shard]

I'd suggest you to just use weights.gg for both images and RVC, easiest way basically

[jaunty shale]

I was literally about to check the training and I got 1% close..

and it disconnected right in front of me.

is there a way to get back??

I use RVC disconnected

[magic musk]

question

INFO:rawvocals:saving final ckpt:Success.

It has finished, but it still isnt overtraining

do i click on train again?

or should i change anything before clicking

[brittle wing]

hi. im trying to run a Google Colab server and for whatever reason its finding errors. i would post screenshots but seemingly have no permissions. anyone might have a fix?
pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts - which it then goes on to list the conflicts.

as a result ngrok server is displaying
ModuleNotFoundError: No module named 'pyngrok'

[low shard]

Could you send your google colab link?

[brittle wing]

hey! i am actually in contact with another member of your team at the moment. if we cant find a fix ill be happy to share it

[low shard]

Oh alr

[hoary estuary]

ho to know which is the better voice model when there are so much voice model of one voice

epoche , and other thing

[low shard]

that's just training info, doesn't tell how good is the model

the only way is listening to the model samples and using it urself

[edgy monolith]

whats the difference between rmvpe and rmvpe_gpu?

[magic musk]

how do i get over 1000 epoch when training?

[rare goblet]

Ayo? @magic musk level 1 !!!

[magic musk]

every time i write like 2000 or 3000 it just goes back to the 1000 i already did and finishes

[low shard]

Different types of RMVPE Pitch Extraction:

• rmvpe: A Robust Model for Vocal Pitch Estimation in Polyphonic Music, the normal version of the best pitch extraction, it's robust and not sensible to noise
• rmvpe-gpu: Training ONLY, uses your gpu for the feature extraction process, using more gpu so making training faster

[low shard]

there's no need to go over 1k

more epochs don't mean more quality

1k is considered the max as there's usually no improvement over that

look at https://docs.ai-hub.wtf/rvc/resources/epochs-tensorboard/

[broken summit]

Is there a tool i can use to upload an audio file and the software converts it into the voice i downlaoded from here? If so, what is it called???

[candid meteor]

my uvr says "zero division error" what do i do?

[rare goblet]

Ayo? @candid meteor level 1 !!!

[granite falcon]

hello its been months since i last came here, now i see i have no acces to the links of the applio or whatever cuz im not verified it says, why is that? i used to have acces

[edgy monolith]

where can i download tensorboard

[deep wedge]

the software shows a bunch of red error message after i tried converting an audio, what to do

[restive heath]

Uhhh install with pip

[glacial pollen]

pip install tensorboard
Then use the batch ( ideally put it in the folder containing tfevents file

then do it like so

paste the path, spam enter few times and done. should open in browser

[edgy monolith]

ok thanks

[glacial pollen]

np, gluck

[runic oracle]

,,,

[brittle wing]

Best model for reverb removal?

[crude flame]

mel/bs roformer reverb removal on mvsep

you can also try clarity vx pro

[brittle wing]

Not on X-Minus???

Okay what settings

[crude flame]

for bs roformer use extract vocals and for mel v2 use as is

[brittle wing]

Not the xminus one?

[crude flame]

xminus doesnt have bs or mel reverb removal

[brittle wing]

Okay I'm which case do I use Mel and which one bs?

[crude flame]

both work well

try both and see which you like more

[brittle wing]

...

It does

[crude flame]

thats not bs roformer

[brittle wing]

Isn't it better?

[crude flame]

bs/mel is better

[brittle wing]

Cause the bs roformer one sounds like it removed more I tried it!

[brittle wing]

@crude flame lastly for denoising?

[crude flame]

There are a ton of denoisers so imma just list them off

• UVR denoise
• Mel denoise
• Bandit V2
• RX spectral denoise
• RX 11's dialogue isolate
• Mel duality V1 can sometimes remove noise in my experience

[brittle wing]

For echo I use the deecho on UVRONLINE

[crude flame]

i think im forgetting a couple but idk

[brittle wing]

Is UVr denoise ok?

Cause it removes noise completely now after using these initial steps

[crude flame]

thats fine but its its small natural room echo you can leave it in

[crude flame]

its alright

also your dataset doesnt have to be 100% noise free it can have some quiet noise

[brittle wing]

Well the previous one had a quiet noise

[crude flame]

quiet meaning like less then 60db

[brittle wing]

Someone suggested these settings... @/Mantrax

[crude flame]

thats good

you dont really need tta tho

[brittle wing]

Will that work

[brittle wing]

I always use it

[crude flame]

most times you wont be able to hear a difference

[brittle wing]

I use it to make sure

[crude flame]

ok

[brittle wing]

Yeah

[brittle wing]

Which Mel V1 or 2?

[crude flame]

for reverb use v2 for duality use v1

[brittle wing]

Huh but BSRoformer derverb takes too many of the vocals away w the reverb itself it's not okay

Are you sure BSRoformer isn't use as is in lead vocals?

I dereverb lead vocals, not full Acapella

MDX23xc doesn't take away that much from the vocals...

It's running the vocals' fullness and quality BSRoformer derverb

[frank owl]

Check out this creation I made on Weights.gg!

[knotty moth]

not sure but it prob means anvuew's melrofo dereverb, and the "v2" is the aggressive one. if you can't find it in the MSST colab, u can try my tweaked version:
https://colab.research.google.com/drive/1IC6Q1hLF55_tK6mhky0SWYKGVF9T5WsY?usp=drive_link

[brittle wing]

What do you use for Acapella?

And what settings for dataset?

What models do you use

For everything

Cause I'm confused

They told me to use Mel-roformer duality 1

But the sound isn't very full on the lead vkcals

Also how much overlap and chunk size do you recommend

How do you make a good dataset?

[knotty moth]

I'm not 100% sure if "they" meant the duality 1 alone directly on the mixture. my thought on it is it has more "fullness" (less muddy) on both vocals and inst than kim's, but also the noise level.

[brittle wing]

Wait how do you make a dataset

[lavish lintel]

So I want to make polish ai model and the polish signs sounds kinda bad do I need to change or use any other pertained model for training. (it will be my first model I don't know how that stuff works so probably best if you can simplify your answers)

[strange brook]

Greetings, comrades! I've been looking into this RVC of ours and really liked how it works and sounds in various demos, so far the best way to copy voice with a seized means of production(locally). But having to have so many of various options to consider I question you if there is a full guide for RVC to be in 2024?

[knotty moth]

there's no pretrain in polish language but you can use more dataset length up to around 30 mins with enough variety and the default pretrain should be fine

[lavish lintel]

If I used 6h dataset how many epochs would I need what is like average or is there any way to check when it's starting to overtrain and is it so bad to overtrain model?

[azure marsh]

Epoch: The number of iterations performed to complete one full cycle of the dataset during training. It's not possible to say precisely how many epochs you need for your dataset, you need to monitor the TensorBoard Graph to know if your model is overtraining.

[bitter lynx]

how do i get rid of the command prob window that opens with the rvc

[tame mica]

you dont

[sour torrent]

does someone knows where to find enhypen ai voice models?

[rare goblet]

Ayo? @sour torrent level 1 !!!

[frigid sparrow]

hi all! Can I train a model on some online service like Google Collab? If yes, please share the link

I have an AMD processor that does not support training((

[restive heath]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[restive heath]

Applio , rvc disconnected , rvc mainline pick what you like
or try kaggle by do -kaggle command

[barren acorn]

so what version should i download voice changer gtx 1660?(6gb?)

[strange brook]

Greetings, comrades! I've been looking into this RVC of ours and really liked how it works and sounds in various demos, so far the best way to copy voice with a seized means of production(locally). But having to have so many of various options to consider I question you if there is a full guide for RVC to be in 2024?

[rare goblet]

Ayo? @strange brook level 1 !!!

[polar tendon]

I did say i was meaning colab by installing

And also i do have 8 gb of ram on that PC

[low shard]

realtime voice changer for calls? wokada, especially the fork

-rt

[azure marsh]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[low shard]

1st link

[low shard]

I mean technically u could do local CPU inference

but i'd suggest cloud for better speed

[low shard]

well there's the docs https://docs.ai-hub.wtf but needs to be updated on some things

[polar tendon]

I would do that if my option i found worked but oh well ¯⁠\⁠_⁠(⁠ツ⁠)⁠_⁠/⁠¯

Time to spend like 7 minutes installing applio on colab which is my only other option

[low shard]

There is also weights.gg and Ilaria RVC Zero if u want

[barren acorn]

yes

[strange brook]

ahh, I knew there would be some kind of doc there. Thanks you for sharing it with a simple proleterian comrade!

[polar tendon]

Well Ilaria RVC does have that problem with uploading models from the device itself

[low shard]

Did you upload the index too?

U need both added index and pth

If u don't have the index, u can just upload any model index and then out the index ratio to 0 in the inference settinga

[polar tendon]

So i can just add a blank file? Also not even applio is safe from that

[low shard]

Not really, i thought that the issue was about you not having the added index but it seems it's smt else

could u share where u got the model?

[polar tendon]

I trained that model

Using RVC Disconnected

[low shard]

could u share the zip? u can just upload it to huggingface so I can use it, or send the .zip on disocrd but i feel like it will be too big

[polar tendon]

Lemme upload it rq

[low shard]

alr, if u dunno u can read https://docs.ai-hub.wtf/essentials/voice-models/#uploading-to-hugging-face

[polar tendon]

I uploaded it here, both zip and non-zip version

https://huggingface.co/RandomPersonRR/Not-even-sure-if-its-a-properly-trained-model-of-me/tree/main

[low shard]

btw u forgot to train the added index

which is very important as it's the accent of the voice

[polar tendon]

It errored out when i tried it

It said it didn't find something about feature extraction which made no sense because i did run it

[low shard]

I think it's the no-feature-todo, meaning that there's something wrong in ur dataset

what program did you use to make it?

[polar tendon]

RVC v2 Disconnected on colab

[low shard]

and be sure have .wav/.flac files, not other types

[low shard]

I mean your dataset

[polar tendon]

I did have .aac files

To save space

[knotty moth]

in case of no-feature-todo, try re-export the wav files in audacity, if you can't open using it try another audio editor to re-export without including metadata

[knotty moth]

ideally avoid lossy formats, the lossless flac should be fine

[low shard]

.aac? u shouldn't have those

btw yea ur model is cooked

u have to retrain it

[polar tendon]

That's fine by me, only took like 10 minutes, time to batch convert them to wav while I'm at it

[rare goblet]

Ayo? @polar tendon level 5 !!!

[low shard]

Reminder for your dataset:

• Have only .wav and .flac , no other types bc only those are lossless (they don't lose data,quality)
• If it gives a no-feature-todo, you can import your dataset in audacity, then export it as a wav

[low shard]

What app did you use to make your dataset btw?

cus i feel like you used smt like capcut

[polar tendon]

You mean to put them in a zip?

[low shard]

No

did you put all the audio files in a single audio or just left it in multiple ones?

[polar tendon]

I don't think a video editing app would export audio, i used lexis audio editor on android to record

[low shard]

oh right you're training your voice

[polar tendon]

I used three of one at different compressions

Lemme check actually it was a bit since i last trained it

Three of one basically

[low shard]

Lemme recap you:

1. Record yourself, be sure that you're in a quiet room and have as less noises as possible, you can say things for like 5 mins. The dataset should be high quality
2. Be sure that the audio file is exported as a .wav or .flac, not sure if the app you're using does that tho, maybe it's in the settings
3. You can have a single audio file or multiple, that shouldn't matter

[low shard]

yea it's bad, you need high quality to train AI

also, be sure to use the tensorboard

[polar tendon]

Doesn't rvc disconnected have that?

[low shard]

yes it opens automatically, im saying be sure to check it in case your model overtrains

[polar tendon]

True

But a nearly 10 minute audio file at 200 epochs should be good

Worked with another person

[low shard]

doesn't really matter tbh, there isn't a right amount of epochs

[quasi dagger]

Indeed, it's a good amount of epochs, but as Nick said make sure to check the tensor board graphs just to be sure

[low shard]

It's better to always check tensorboard imo

tensorboard is important

[polar tendon]

Yeah

[quasi dagger]

Yeah, I mean that arbitrary value is a good place to start, but check the graphs

[polar tendon]

I am

[quasi dagger]

Cause some models overtrain earlier

[polar tendon]

What loss rate is a cue to stop tho

[quasi dagger]

When the graph starts rising again and/or no much loss (like a flat line)

[polar tendon]

At any number? Like 4, like if it's flat there?

[quasi dagger]

This is an example of a rising trend, the optimal spot is before it starts rising

[quasi dagger]

No, because each model is different

[queen silo]

how do i train a voice in applio?

[rare goblet]

Ayo? @queen silo level 1 !!!

[queen silo]

i want to do voices from popular shows from brazil

but i have no idea to where to start

like do i have to get an audio or smth like that?

[polar tendon]

Multiple if some are lower quality

[queen silo]

ok

i just dont know where to start

...

[quasi dagger]

Yes, you need to make a dataset which is a collection of audio files you will use to train a voice model

[azure marsh]

Dataset

A set of audio files compressed into a .zip file, used by RVC for voice training. The quality & length of the dataset are the biggest determining factors of the final quality of the model.

[queen silo]

thanks

[polar tendon]

For applio, the dataset creator is where you put all your audio files, once done importing all the audio files, you hit the preprocess button

[queen silo]

ok

is 5 minutes good enough or it needs more?

[polar tendon]

Yeah should be good enough, if it sounds inaccurate once training, i would go with 10 minutes with 200 epochs

[queen silo]

i will try with 40 secs jus for curiosity

[queen silo]

k, this will be very hrd

[polar tendon]

To get right

[queen silo]

yea, but i'll try 5 mins

[rare goblet]

Ayo? @queen silo level 2 !!!

[queen silo]

the video was cut in half for some reason

[polar tendon]

Weird

[queen silo]

but it was prob cause of the converter

i cant find the model

i just found a folder with sliceed audios

[distant turtle]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[polar tendon]

It does slice audios to train it, It's to make it easier for something

@low shard @quasi dagger @knotty moth thanks for the tips and stuff! it helped with my model sounding pretty accurate.

[quasi dagger]

You're welcome!

[untold pivot]

Hey guys, I was wondering if there is a naming convention for index files. I’m using Mangio RVC for my current internship but I keep having a problem when I put in the inferencing voice. It gets caught in a loop of trying to load the index file.
The only problem I can think of is the name of the index file as the only model that was working previously has a weird naming convention

[brisk nova]

im trying to open the RVC but its rapidly closes

what do i do

i already istalled python

but its still closing when i click 2 times on the RVC

[sage ether]

how do i use a voice model?

[brisk nova]

Help pls

[rare goblet]

Ayo? @brisk nova level 1 !!!

[hasty lichen]

help pleas

[odd shale]

!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.

[odd shale]

!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.

[odd shale]

Please you both, explain properly what's the problem you got. 😄

[hasty lichen]

i downloaded ai voice changer from this

https://huggingface.co/wok000/ttsclient000/blob/main/ttsclient_win_cuda_1.0.4.zip

and im so lost how to install it

[odd shale]

Actually.. If you want help regarding W-Okada usage, you can go to the #🔍│help-w-okada channel.

[odd shale]

Btw i can suggest you to use the Deiteris fork instead of that one.

But, what's your gpu?

[hasty lichen]

rtx 3080 ti

[odd shale]

You'll be fine then.

There you have a guide.

https://rentry.co/ForkVoiceChangerGuide

[hasty lichen]

ty bro

[odd shale]

You're welcome bud.

If you want help about W-Okada/Deiteris fork, head over to the #🔍│help-w-okada channel.

[hasty lichen]

but just one more q, does it support text speech with diffrent voice models

[odd shale]

No.

It doesn't support tts.

W-Okada is for realtime usage.

[hasty lichen]

oh nooo

do you know some free app that does 😭

[odd shale]

Ehh.. Just use Applio colab.

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

If you wanna do it quickly.

Or also, you can read the docs about how to install RVC/Applio locally

https://docs.ai-hub.wtf/

[hasty lichen]

so i can get any voice like movie or anime and use it on my text speech

[odd shale]

Actually for using an X character's voice on TTS, you need at least a RVC model of them and use it on Applio.

[hasty lichen]

i think this server got some if im right?

[odd shale]

Yep, it does.

[hasty lichen]

oh you helped me a lot ty dude, im so lost on the site you send me but i will try how to get the app installed lol

[azure marsh]

Hey, abdog_btw! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[odd shale]

You're welcome..

[candid meteor]

my uvr says "zero division error" what do i do?

[azure osprey]

No matter what version I try I literally cannot get this thing to work. NVIDIA GPU(RTX 4060) currently trying to use the version MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a

But every time I even get the thing to run, it either doesn't change my voice, or in this case I'm stuck in a white screen with the loop of
[VCClient] wait web server...
As of writing this message it's been 220 seconds.

[pastel oak]

delete stored_settings.json for wait web fix

changing voice do you have virtual cable installed and setup correctly

[azure osprey]

I'm just trying to change it in the app for now, I havn't even gotten to the point of trying to use it in anything else

[rare goblet]

Ayo? @azure osprey level 1 !!!

[azure osprey]

But lemme try to delete the storage file

Yeah now it just doesn't want to work...

like it doesn't launch

I'll try to restart installation one more time ig

[lime heath]

best settings for gtx? pls

actually which1 do i even download for gtx

im on 1080ti

[brisk nova]

How

[coarse heron]

so which is better, Okada or Deitris. sorry im new to all of this.

nvm i'll just read the descriptions on the links you guys provided.

[odd shale]

Deiteris (i think) tends to be better than Okada.

Technically both are the same, but one got some tweaks and improvements.

[coarse heron]

yeah, i just finished reading the article. thanks man.

[rare goblet]

Ayo? @coarse heron level 1 !!!

[odd shale]

You're welcome buddy.

[coarse heron]

haha yep, lvl dude.

lvl 1

imma use it for voice overs and i dont wanna pay female employees so imma do it myself lol

[coral frigate]

hi is there any way to upsample a model from 32k to 40k? i dont have the original dataset

[odd shale]

No, you can't upsample a model.

But the only way to do it is making a dataset of the same person/character but with higher samplerate.

You can check the samplerate of a file using spek or RX

[brave garnet]

RVC Colabs and Spaces

⠀

HuggingFace Spaces 🤗

⠀
Ilaria RVC
EasyGUI port with some improvements, by Ilaria.

RVC-HFv2
Applio port, by r3gm.

AICoverGen
AICoverGen port, by r3gm.

Advanced RVC Inference
Extended version of the GUI with advanced settings, r3gm.
⠀

Made with love by Ilaria 💖

[simple ore]

nope