✨│ai-help

[nocturne marten]

Does not matter, I just need it to work :)

I'll see if I have any changes

[rare plinth]

okay! lmk if it works or not

[brittle wing]

hey can u help me

everytime i try my voice it crashes

@acoustic scarab

@tired kraken

@rare plinth

@steel forge

@hearty idol

[brittle wing]

does anyone know how to use W-Okada on the macbook?

I downloaded the thing and it doesn't work

[nocturne marten]

I discovered that python was technically never installed on my PC, which is why I installed it and reinstalled Applio and now I get a different error. I will just try to use an older version and hopefully it will be better

Thank you for the help!

[rare plinth]

@nocturne marten no problem! happy to help. :D

[rare plinth]

@brittle wing my apologies for being unable to respond in time. can you show me the model files, perhaps in a direct message if you're uncomfortable with sharing it publicly?

[nocturne marten]

Good news! Version 3.24 is buggy, use 3.2 Instead (at least for windows)

[rare goblet]

Ayo? @nocturne marten level 2 !!!

[rare plinth]

hooray!

[cedar imp]

anyone got any clue to fix the test_suite error for fairseq?

[rare goblet]

Ayo? @cedar imp level 1 !!!

[somber root]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer 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, -overtraining, -realtime, -rvc, -help

[steel forge]

Please don't ping staff in mass, people tend to find it annoying

What happens when it crashes? Does it crash immediately upon startup? Are you getting any feedback in the console window that opens with Okada?

If there's an error, it'll usually show in the console

[weary lance]

how do I install it?

could someone help me

[low shard]

What's your PC GPU?

And install what?

[weary lance]

1660 super and the ai voice changer

[weary lance]

.

[low shard]

GTX 1660 and realtime voice changer for calls right?

[weary lance]

and for games like val

[rare goblet]

Ayo? @weary lance level 1 !!!

[low shard]

Alr

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[low shard]

Read the first link guide

And for any issues, ask in #🔍│help-w-okada or #1192011222023950368

[weary lance]

ok thanks

[low shard]

Yw

[south nest]

ModuleNotFoundError: No module named 'mega'
Can anyone help me fix this? This is my first time doing AI so I don't know much. Please sympathize with me

[carmine hearth]

Which colab do you use?

[south nest]

Ilaria

[rare goblet]

Ayo? @south nest level 1 !!!

[tough fiber]

bruuuh

and yea its 0.987

[frozen iron]

okay thank you for further info, perhaps one day ill have a go 🙏

[pulsar sequoia]

Is there any old captain price audio file? His voice in 2009.

[golden walrus]

dumb question, 😦 i followed the web instruction but i can't install tensor board

is there a guide how to install it ?

oh nvm, i found it

[rare goblet]

Ayo? @golden walrus level 3 !!!

[proven hill]

me!

[low shard]

You seem to be using the old Ilaria RVC mangio Google colab #✨│ai-help message

Just use Ilaria RVC Zero for inference (using models)

[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! 💖

[scarlet wedge]

how to fix this error?

[neon thicket]

what should I do if I hear myself all the time?what should I do if I hear myself all the time?

I turned off the program and pressed stop, but I can still hear my voice.

[red cliff]

Can you train the index before the actual model or is that not good

[pastel oak]

Doesnt matter when, most do it before since its just a few seconds

[pastel oak]

That does not seem to be related to voice changer, you have a different program like voicemod maybe thats accessing virtual cable or something

[knotty moth]

it can be before or after training actual model, but must be done after preprocessing and feature

[rare goblet]

Ayo? @knotty moth level 33 !!!

[neon thicket]

but do you know how remove this?

[scarlet wedge]

any colab for inference by uploading my own voice model without need to use link from huggingface?

[low shard]

you can use applio colab

but i'd suggest you to use Ilaria RVC Zero

its a ZeroGPU hugging face space, which runs on A100 which is 11 times faster than google colab T4

[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! 💖

[scarlet wedge]

thanks

i need to make private inference, my model private

[low shard]

in ilaria rvc zero only the model is public

but i understand

[low shard]

yw

[golden walrus]

dumb question, i trained an ai, but then i only use 2 small file in weight and in the logs file? what should i do with these D and G files

[proven hill]

throw them in the bin

[golden walrus]

so just delete it ? okay, on my wayyyy

[winter panther]

is there any free site to train ai voice?

[low shard]

there are the cloud ways yea

As you dont got a good PC, its better you use cloud 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)

[mystic tangle]

where do i find the pth and index i let illaria rvc train it on colab then it disconnected idk when am i still able to find whatever it trained?

or is it all gone

[low shard]

yea, the G & D are files used as pretrains soo u don't need them unless when u train a pretrain

[golden walrus]

so in another words, it's no use when i got the voice model ?

[low shard]

exactly

[golden walrus]

thank you

but i met a problem where the model i trained made weird noise or mispronounce. is there anyway to improve it? or i need a better data ?

[odd shale]

Probably those issues are because you didn't clean properly your dataset, your model did overtrain or the sample you used to test with your model got noise too.

But yeah, that's fixable if you clean better the dataset and retrain.

And also making sure to not overtrain the model.

[golden walrus]

i trimmed and did everything i could but the audio from the game is not that clean

[odd shale]

Hmm.. That explains it.

[golden walrus]

ah i forgot to say, i can't stop training process

they told me to press stop training where the training button were

but i clicked and nothing happened

[odd shale]

Are you using colab, local or kaggle?

[golden walrus]

i'm using local

[odd shale]

Alright then.

[golden walrus]

but i get the hang of things now. thank you

[earnest knoll]

so I've been training a model using Applio (locally), and it finally finished with training and it created the final .pth file, but it failed to generate the index and I can't find a way to start generating the index without restarting the training process all over again. What do I do?

[thin stump]

you can always do the index before and after so just insert the same model name, target sample rate, pretrain batch size, save frequency and train the feature index

without restarting
all your models are saved in the logs folder

[brave garnet]

RVC Colabs and Spaces

⠀

Google Colabs

⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.

AICoverGen-NoWebUI
Useful for making covers, doesn't include a UI, by Ardha, by Eddy, Hina and Gdr.

RVC Disconnected
To train new voice models, by Kit Lemonfoot.

EasyGUI
The OG interface, by Rejects.
⠀

Made with love by Ilaria 💖

[floral sleet]

i can't use rvc

pipeline

[pastel oak]

Delete pretrain and model_dir, then try starting the program again

[floral sleet]

ok

[brittle wing]

-colabs

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer 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, -overtraining, -realtime, -rvc, -help

[brittle wing]

-hf

[azure marsh]

<:huggingface:1179800228946268270> Hugginface Spaces

• UVR5 UI, by Eddy and Ilaria Huggingface Spaces
• Ilaria RVC Zero, by Ilaria Huggingface Spaces
• RVC⚡ZERO, by r3gm Huggingface Spaces
• Applio, by IA Hispano Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -colabs, -kaggle, -local, -overtraining, -realtime, -rvc, -help

[brittle wing]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[brittle wing]

-help

[azure marsh]

Command has expired

[rare goblet]

Ayo? @brittle wing level 3 !!!

[brittle wing]

-help

[azure marsh]

Command has expired

[brittle wing]

-guides

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[radiant bison]

whats the website called

for the rvc

ik theres a site

@low shard you told me it last time i frogot what its called

[low shard]

idk if we talked before as u joined just today

So, you want to inference (use model) or train (make model) ?

[distant condor]

anyone help me to bulit voice clone

[night relic]

Hello everybody,
Anyone an ESRGAN expert here?? I want to get help out in ERSGAN google collab.

[gusty kestrel]

@copper gust, I have found 1 results that match your search!

megaaziib

Vesta Zeta Hololive ID

• Uploaded: <t:1699210306:d>
• Server: Unknown
• Likes: 0
• Lang: English

Epochs

1000

Technology

RVC

Algorithm

Rmvpe

Requested by farnaldo7 · Aug 24, 2024, 8:59 AM

@copper gust, I have found 3 results that match your search!

megaaziib

Vestia Zeta (Hololive ID) (Ov2 Pretrain)

• Uploaded: <t:1704769904:d>
• Server: AI Hub
• Likes: 0
• Lang: Unknown

Epochs

300

Technology

RVC

Algorithm

Rmvpe

Requested by farnaldo7 · Aug 24, 2024, 9:01 AM

@copper gust, I have found 7 results that match your search!

makiligon

JP Furina Focalors CV: Inori Minase RETRAINED (Genshin Impact)

• Uploaded: <t:1700456285:d>
• Server: AI HUB
• Likes: 0
• Lang: Japanese

Epochs

170

Technology

RVC

Algorithm

Rmvpe

Requested by farnaldo7 · Aug 24, 2024, 9:04 AM

[tame mica]

@copper gust #🔍│find-models

[gusty kestrel]

@copper gust, I have found 1 results that match your search!

riesudesuu

Firefly (JP) (Retrained) | Honkai: Star Rail | SnowieV3

• Uploaded: <t:1719323456:d>
• Server: AI HUB
• Likes: 0
• Lang: Japanese

Epochs

500

Technology

RVC

Algorithm

Rmvpe

Requested by farnaldo7 · Aug 24, 2024, 9:19 AM

[upper compass]

I got my GPU working with RVC based projects, but there is some strange problem, if i generate TTS, it takes longer then CPU, if i generate same text again, it takes half second, even if text is big. I can change only one symbol and now it generate audio again longer then CPU, but GPU load is 100% and no errors, i have RX 7800 XT, i've tried different Torch-ROCm version (5.6, 5.7, 6.1) but nothing change

[wispy lodge]

Which RVC-based project you run and on which OS?

[upper compass]

"Retrieval-based-Voice-Conversion-WebUI", Nobara Linux (Fedora Linux 40)

[wispy lodge]

Then not sure, it might be that rvc-webui reloads the model between conversions, and it has to perform the optimization for different input once again. Or it's just what ROCm does under the hood

The thing is that when the input length changes, the framework (pytorch, onnx, whatever) optimizes execution for it. So the first run with this input can be longer than usual

[upper compass]

I understand that, but it's every time slower then in CPU (AMD 7600X)

[wispy lodge]

Well, because first-time optimization pass takes time, and in some cases it can be longer than a run on CPU. Subsequent runs, as you've already noticed, are much faster

But I cannot tell if that's exactly the cause since I don't really work on rvc-webui

[upper compass]

Subsequent runs of same exact text is much faster, but this doesn't make any sense, because text is always different

[wispy lodge]

But how about text that has the same length as the previous one?

[tame mica]

🐠

[upper compass]

Thats not helping 😦
I change few words, but it's same length
Use different models, voices and texts

[rare goblet]

Ayo? @upper compass level 2 !!!

[obsidian scaffold]

Hi anyone know how I can run an inference from cmd or python module, without the gui?

[upper compass]

If this inference gui has Gradio library in python script, then you can use gradio api, it can be founded in the bottom of webpage, research and test it, then try to start inference with arg "--noopen" or something like that

[wispy lodge]

Then something might be wrong in the code or doesn't behave with ROCm as expected

[low shard]

@copper gust #🔍│find-models

[rare goblet]

Ayo? @low shard level 100 !!!

[azure marsh]

💻 (Local) Realtime Voice Changer

• 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

[upper compass]

Is there any way to disable this first time optimization?

[lethal basin]

@acoustic scarab

Help I need

@acoustic scarab

I need help

@acoustic scarab

guys the helper

Won’t help

What is this scam

@acoustic scarab

L help

[brisk quarry]

can someone help me remember how to train

i did it several times successfully but for some reason when i stop for a while i forget how to do it

which gui was the best

@wispy lodgepinging u since ur the only person helping i see when scrolling up

[low shard]

don't spam ping helpers, they aren't robots

[low shard]

what pc gpu do u have?

[brisk quarry]

i meant the steps with like the gui and the folders

i dont even remember what gui i used

but i did a model before

[low shard]

yea, what pc gpu do u have?

[brisk quarry]

🗿

[low shard]

if u are talking about easygui google colab

its broken

[brisk quarry]

no

local

with python n stuff

the webui

[low shard]

yea i asked your pc gpu bc u need a good one for local

[brisk quarry]

i have a good one trust me

[low shard]

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

[brisk quarry]

almost the best components my pc can do 🗿

i know what a gpu is

i have everything and everything is good enough i just need help remembering what i even did to train

like the installation steps

[low shard]

if you say so, asked just to make sure

https://docs.aihub.wtf you can find here mainline (original rvc) and applio (rvc fork with some extra features, same quality tho), u can choose one to download and there will be the guide for training in that site

[brisk quarry]

was applio the good one

i think i used that

[low shard]

kinda weird, the dataset was all clean and in a different accent right?

[low shard]

they both are good, its really ur choice

[brisk quarry]

why is it a .exe now

[low shard]

which one are u downloading

[brisk quarry]

applio

i used zip before

did i miss something

[low shard]

how long was that dataset?

[lethal basin]

Make them gr

[low shard]

i cant do things locally myself, but as seen here u should unzip it and have a .bat

in the hf repo there are the .zip versions

[brisk quarry]

i meant why is there also a exe

[low shard]

they arent really forced to help, they can also be busy, u didnt even say what help u need, and dont spam ping

[low shard]

i think its a .exe installer

[brisk quarry]

was there a specific folder name for the training samples

and files

[lethal basin]

Butt I need Alexeys help

[low shard]

16 seems big, however what pitch extractor did u use? rmvpe right?

[low shard]

there are other helpers too, if u continusly spam ping people u could get muted

[wispy lodge]

I think that's possible only in the code

[lethal basin]

ok

[lethal basin]

When does Alexey help me

I’ve been waiting

[brisk quarry]

.

[knotty moth]

why u simp for him?

[brisk quarry]

[lethal basin]

Wtf !!

I like annoying him

@acoustic scarab

Hello

Helpppp

Meeee

[low shard]

the weights and index? i think logs folder

[low shard]

yea, no.

[brisk quarry]

i forgot what index is

also why cant i send images

[low shard]

spam ping = mute

[lethal basin]

Trust I need help

[lethal basin]

Ok

[brisk quarry]

what batchsize should i use for 12gb

[fleet cedar]

why does "mon" doesnt show for me

[brisk quarry]

?

[low shard]

need level 1 to send images in help channels

rmvpe is usually the best

[low shard]

the file of the accent of the voice

[low shard]

u looking for a voice model?

[brisk quarry]

hmmm

[knotty moth]

I suppose "monitor" in the voice changer?

[brisk quarry]

how does tensorboard work again

can i use a sample that is like 5mins long

1 file

:/

what batch size do i use for 12gb

Traceback (most recent call last):
File "C:\Users\Lukag\Desktop\ApplioV3.2.4\env\lib\multiprocessing\process.py", line 315, in _bootstrap
self.run()
File "C:\Users\Lukag\Desktop\ApplioV3.2.4\env\lib\multiprocessing\process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "C:\Users\Lukag\Desktop\ApplioV3.2.4\rvc\train\train.py", line 449, in run
train_and_evaluate(
File "C:\Users\Lukag\Desktop\ApplioV3.2.4\rvc\train\train.py", line 673, in train_and_evaluate
wave = commons.slice_segments(
File "C:\Users\Lukag\Desktop\ApplioV3.2.4\rvc\lib\algorithm\commons.py", line 73, in slice_segments
ret[i] = x[i, :, idx_str:idx_end]
RuntimeError: The expanded size of the tensor (12800) must match the existing size (0) at non-singleton dimension 1. Target sizes: [1, 12800]. Tensor sizes: [0]

@knotty mothcan you tell

what batch size

[pastel oak]

Are u not the guy that ragequitted the rvc server because you didnt get a reply in 1 hour lmao

[brisk quarry]

yes

[pastel oak]

Itd be good to tell us if you have an NVIDIA RTX or not instead of saying "trust me" to see if the gpu is the problem or not lol

[brisk quarry]

and

[pastel oak]

batch size 8 is fine usually

Tensorboard is explained on the guide:
https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/

[brisk quarry]

is it even correct the gui is saying higher is faster

i heard once higher is slower and use more vram

[pastel oak]

higher is faster and uses more vram

[brisk quarry]

eh

and what is more accurate

[pastel oak]

But with insufficient amount of dataset length, it can cause issues

I would stick to 8

[brisk quarry]

.

i have enough time

even if its slower

[brisk quarry]

is 5-6 mins long enough

[pastel oak]

8

[brisk quarry]

what

what 8

[pastel oak]

Batch size 8

[brisk quarry]

i asked if its long enough

or too short

[pastel oak]

10-20 is better

Some models can be fine on 5-6

[brisk quarry]

you still didnt tell me what is more accurate

[pastel oak]

You didnt tell me accurate in regards to what, so I assumed u mean batch size which I said 8 4 times

[brisk quarry]

i have no idea either

the gui says 4 is more accurate

maybe in voice accuracy

[pastel oak]

Since you have a lot of time, train both with 4 and 8 and see for yourself then

[brisk quarry]

how to with both

[pastel oak]

You train one with batch size 4

You finish training, then train it again with batch size 8

[brisk quarry]

so higher is better?

[pastel oak]

Your sentence has no correlation to the graph

You select the one with the lowest point in g/total after some time of training

This point can be anywhere, every voice model training is unique, so theres no definite number

[odd shale]

Nope, take a look at the lowest point on that graph.

[brisk quarry]

the site says the marked point is when its overtrained

[odd shale]

Go for the .pth file that matches or it's near that lowest point.

[odd shale]

Yep, because after that lowest point, it's overtraining.

[brisk quarry]

so im supposed to keep it low

[pastel oak]

You dont do anything, it does it automatically, u then select the lowest

[brisk quarry]

wdym select

so overtraining is just when its running for too long?

[odd shale]

Not necessarily.

In few words it means it already converged.

So, you must grab the lowest point on the graph.

[brisk quarry]

what can i do if the voice result is too high pitched

[odd shale]

what do you mean with too high?

[brisk quarry]

its supposed to be deep

[odd shale]

Ehh.. I'm not sure but in that case it will already depend on the audio you use to test with your model.

And also the pitch.

Normally when you start rvc, on the inference tab, the pitch will be always in 0.

[brisk quarry]

wdym "also"

do i lower it

[odd shale]

In few words, yes.

[brisk quarry]

i have that one pitch setting enabled idk whats it in english my gui isnt for some reason

pitch height support or sum

[odd shale]

Depending on the voice you trained you must lower or put the pitch higher.

Transpose is pitch on some RVC versions.

[pastel oak]

Just start with training your model, make sure you select save frequency 15, save every model per frequency and look at tensorboard. If you notice the graph keeps going up after some time of training, then check what "step" number the lowest point in the graph has (example: 11280) then go into your weights folder, check for model_name_eXXX_step_11200, whatever is the number before the 11280 and you do testing with that model

For the "what can i do if the voice result is too high pitched" you lower the "transponse"/"pitch" to the minus area if 0 is too hgih pitch

[brisk quarry]

i dont see a setting for changing the pitch

[odd shale]

Go to the inference tab.

[brisk quarry]

im on training tab

[odd shale]

And look for something that says "transpose"

[odd shale]

The model is still training?

[brisk quarry]

it finished

but i wanna train it lower

is inference also training?

[pastel oak]

Training has nothing to do with pitch

Inference tab is making AI Cover

[brisk quarry]

then what is the model for

[pastel oak]

So if you are ready to test the model, and you want to change the pitch, you lower the TRANSPONSE setting

You need a model to make an ai cover

[odd shale]

Nope, that's not possible. Training doesn't have anything to do with the pitch.

[pastel oak]

You have an audio of idfk dua lipa singing

You have a model of Joe Biden, you insert the audio and select Joe Biden model, and you make Joe Biden sing that song

And if the pitch is too high, you lower the transponse on the Inference Tab which you make the covers with

Does that make sense

[brisk quarry]

then what even is the model for

[pastel oak]

I give up

[brisk quarry]

you never really answered it

you need a model isnt a real answer

what does the model do

[odd shale]

A RVC model converts an audio you selected to sound like the person you made a model of.

That's it

[brisk quarry]

and why not the pitch

[odd shale]

On some RVC GUI's, transpose is pitch.

[brisk quarry]

how do i change my guis language

nvm

[odd shale]

I don't even know, sorry.

[brisk quarry]

its in the settings

[odd shale]

🐢

[brisk quarry]

is more files better or does only the duration matter

what does that do on training

[odd shale]

Makes the model able to sing in few words.

If you disable that the model after training when testing it, will sound monotone

[brisk quarry]

like a npc

[odd shale]

Yep, without that enabled it will sound like a npc/monotone

Keep it enabled.

[brisk quarry]

is that the right graph

[knotty moth]

different sample rate or some configurations than set before

[knotty moth]

without pitch guidance means no pitch extraction

[brisk quarry]

why does the graph spike so hard

[fleet cedar]

yes

[low shard]

You can search rvc ai voice model:

• #1175430844685484042
• #🔍│find-models
• https://weights.gg/
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models
• https://voice-models.com/

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.aihub.wtf/essentials/how-to-make-voice-models/

[fleet cedar]

but my accent is so shit

[chrome sequoia]

I dont know if this is the right # but I would love to get some advice/help to get started on my speaking human avatar project that can present a text segment, 1-2 minutes long. Any advice on script,model,code is useful :))

[odd shale]

Speaking human avatar?

[brisk quarry]

does the duration of 1 video matter more or the amount of videos

[brisk quarry]

what is :\Users\Lukag\Desktop\ApplioV3.2.4\env\lib\site-packages\torch\autograd_init_.py:251: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed. This is not an error, but may impair performance.
grad.sizes() = [64, 1, 4], strides() = [4, 1, 1]
bucket_view.sizes() = [64, 1, 4], strides() = [4, 4, 1] (Triggered internally at C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\reducer.cpp:334.)
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass****

@odd shale

[brisk quarry]

[brisk quarry]

why is vrc training not using all my vram

[tidal trail]

if you are using applio in settings tab

[tidal trail]

it's normal

[brisk quarry]

but why is vrc training not using all my vram

only like the half

[tidal trail]

sorry i do not train locally, i think the batch size that you choose

[brisk quarry]

higher?

[tidal trail]

set in 8 or 4

[brisk quarry]

its already 4

[knotty moth]

batch size 8 takes 8 gb vram, less than that takes less

[brisk quarry]

so i set to 12 or what

also i set to 4 and its using 8gb

so what are you yapping about

[knotty moth]

some explanations of relativity theory and time dilation, and quantum computing

[brisk quarry]

🤓

[brisk quarry]

ong i set to batch size 14

better gpu

faster

[rare goblet]

Ayo? @rapid falcon level 1 !!!

[brisk quarry]

idk

reinstall

[vague magnet]

Hi guys, I'm new to this. Why does my voice lag so much with the sounds I added, except for the ones provided by the program

[grizzled cove]

Why does the AI struggle so much with this song? Skip forwards to ~0:20 for the actual sounds. It had little issues with other music, but for some reason it always struggles with songs like this.

[pastel oak]

What about it do you not like?

And what separation method and model are you using

[grizzled cove]

For separation I used 🎵 UVR5 NO UI 🎵 with BS-Roformer and BS-Roformer-Viperx-1297.ckpt
For the AI cover I'm using Ilaria RVC 💖

[pastel oak]

Wait you need to be more specific. What do you not like about it? Do you mean the AI covers are the problem, or do you not like the separation from the instrumental

If you mean ai covers - yeah, this song has a shit ton of reverb. You would need to de-reverb it before making a cover of it

Also, it is a double-layered voice (zwei/mehrspurig), so it will always struggle because its hearing multiple voices on top of each other - not clean enough

[grizzled cove]

Here's the cover, it's... pretty bad imo.

[pastel oak]

Yes, this is because of the multi-layered vocals

[odd shale]

Did you make sure to remove the layered/backing vocals and reverb/echo?

Also, in this case i would suggest you to look for a cover of that song to use it instead too

Sometimes covers of songs made by random people don't even have backing vocals neither harmonies.

[brisk quarry]

Bruhhh

Because its german

[turbid owl]

im trying to start the easygui and this happens and i dont know what to do

[rare goblet]

Ayo? @turbid owl level 2 !!!

[low shard]

Easygui google colab is broken #📰│dev-updates message

[jaunty glade]

what does 'crepe hop' mean
im fiddling with crepe-tiny on that one RVC GUI ._. XD like is it for audio 'pitch depth' ?

[jaunty glade]

aaaah Thank you! for information

[rare goblet]

Ayo? @jaunty glade level 1 !!!

[jaunty glade]

I WILL LEAVE YOUR BODY IN OKLAHOMA

[summer estuary]

hello

My audio file is currently over 5 minutes and can't use Medleyvox, can you guys help me

[odd shale]

We don't know about MedleyVox colab, sorry..

But you can try models like BVE and/or Mel Karaoke on MVSEP.

[turbid owl]

how has it been broken for a month

shouldnt they hav efixed it by now

[low shard]

nope, creator is inactive since may

a 'fix' would be putting !pip install pip==23.1 at the top of the code and run it, but i checked the code and there could be other outdated things, i highly suggest to just use an alternative

as the google colab will probably never get updated and stay broken

[turbid owl]

ok

ilaria rvc is just giving me errors too

[brisk quarry]

use applio

[low shard]

are you using the mainline google colab?

u should show the error in the google colab page

also, there is a easier way, ilaria rvc Zero

[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! 💖

[brisk quarry]

applio

[uncut yoke]

my discord is picking up my voice, or the voice changer

[rare goblet]

Ayo? @uncut yoke level 2 !!!

[uncut yoke]

i have everything set up correctly but i cant hear my voice in a call with my alt

[gusty kestrel]

@marsh nebula, I have found 1 results that match your search!

fellipecaveira

Yuji Itadori (Jujutsu Kaisen) (Yuri Tupper)

• Uploaded: <t:1702236984:d>
• Server: Unknown
• Likes: 0
• Lang: Japanese

Epochs

350

Technology

RVC

Algorithm

Rmvpe

Requested by galunki · Aug 25, 2024, 12:12 AM

[knotty moth]

try BVE v2 or mel roformer karaoke in x-minus (need pro subscription), or download the latter for UVR whose link can be found in
https://docs.google.com/document/d/17fjNvJzj8ZGSer7c7OFe_CNfUKbAxEh_OBv94ZdRG5c/edit

[knotty moth]

-gui

[azure marsh]

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

[knotty moth]

you should use one of the recent ones:

• Ilaria RVC
• mainline RVC
• Applio

[astral perch]

I noticed that 3 audios in my dataset are way louder than the other audios. Will this have a really bad impact on the quality of the model?

[brisk quarry]

https://tenor.com/view/yoinky-sploinky-yoinky-sploinky-meme-dancing-gif-20769405

[chrome sequoia]

Yeah like deepfake, or just a picture of someone that can be made into engaging talking videos

[ivory forum]

why are some on my F0 det options not available (n/a) like crepe full or harvest?

and also if i have the onnx version of something is it better than with one with index where the voice model is trained?

[tame mica]

https://tenor.com/view/mayonnaise-cat-furry-beecat-meow-gif-26509467

[rare goblet]

Ayo? @tame mica level 71 !!!

[obsidian spire]

Hi, could anyone help me? I've been trying to create covers in Applio on Google Colab, but I get this.

File "/usr/local/lib/python3.10/dist-packages/gradio/queueing.py", line 532, in process_events
response = await route_utils.call_process_api(
File "/usr/local/lib/python3.10/dist-packages/gradio/route_utils.py", line 276, in call_process_api
output = await app.get_blocks().process_api(
File "/usr/local/lib/python3.10/dist-packages/gradio/blocks.py", line 1923, in process_api
result = await self.call_function(
File "/usr/local/lib/python3.10/dist-packages/gradio/blocks.py", line 1509, in call_function
prediction = await anyio.to_thread.run_sync(
File "/usr/local/lib/python3.10/dist-packages/anyio/to_thread.py", line 33, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "/usr/local/lib/python3.10/dist-packages/anyio/_backends/_asyncio.py", line 877, in run_sync_in_worker_thread
return await future
File "/usr/local/lib/python3.10/dist-packages/anyio/_backends/_asyncio.py", line 807, in run
result = context.run(func, *args)
File "/usr/local/lib/python3.10/dist-packages/gradio/utils.py", line 832, in wrapper
response = f(*args, **kwargs)
File "/content/program_ml/core.py", line 76, in run_infer_script
infer_pipeline.convert_audio(
File "/content/program_ml/rvc/infer/infer.py", line 173, in convert_audio
self.get_vc(model_path, sid)
File "/content/program_ml/rvc/infer/infer.py", line 322, in get_vc
self.setup_network()
File "/content/program_ml/rvc/infer/infer.py", line 358, in setup_network
self.tgt_sr = self.cpt["config"][-1]
KeyError: 'config'

Does anyone know what it means?

[plush compass]

Is there a link to UVR Colab?

[brisk quarry]

why does it not make a index file when i click

[ivory solar]

Best thing I heard in a while

[pastel oak]

Delete if this has copyrighted instrumental in it

[brisk quarry]

it dont

its a edited rap

i think

idk

i dont even know from weho it is

so ig

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer 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, -overtraining, -realtime, -rvc, -help

[brisk quarry]

how do i train the pitch also for lower voices cuz its too high

[brisk quarry]

why is this channel so dead

[odd shale]

Nep, we've already told you training has nothing to do with pitch

You must change the pitch depending on the audio you use to test the model with.

[brisk quarry]

then whats the point if it dont apply the pitch

[raven viper]

hi

I come from Vietnam

[brisk quarry]

ok

[odd shale]

Transpose is pitch.

[raven viper]

i don't know how to download anime voice changer

can you help me

[odd shale]

You must change the transpose depending on the voice you trained.

[odd shale]

There you go buddy.

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[odd shale]

It's called W-Okada, not Anime voice changer btw 😄

[raven viper]

can you private message me

[raven viper]

ohh okay

[brisk quarry]

and what is that.

i want the voice to work with default settings and not having to change the pitch

[odd shale]

Sorry but that won't be possible.

That's not how RVC works.

[brisk quarry]

but i did markiplier with deep pitch before

[odd shale]

Because you probably used an audio that matched Markiplier's voice.

That's why.

Depending on the audio you use to test with your models, you must forcefully change the transpose/pitch to make it sound like the person.

I don't wanna be rude but... To this point you're just refusing to understand for no reason.

[brisk quarry]

im trying to understand why

also i asked several times and nobody respond but is it better to use more short files for training or less and longer files

[odd shale]

Umm.. I was talking about model pitch, not dataset lenght but if you want an answer, it depends

[brisk quarry]

on?

[odd shale]

Sometimes the dataset lenght doesn't really matter.

But instead how clean it is.

For making a proper model you need a decently clean dataset of any lenght. You can use any dataset length for your model.

[pastel oak]

Example:

You have markiplier voice model

Audio you put in is woman -> Markiplier voice model copies the frequency and pitch range because thats what it hears, it copies the pitch range of the audio you put in automatically, thats just how it works. So if you want lower pitch back to deep voice, you lower pitch, aka. you manipulate the pitch

Was that helpful?

[brisk quarry]

and what if the pitch varies

[pastel oak]

RVC usually splits up audio itself so in the end it doesnt matter too much, but I would still encourage you use many, shorter files

[odd shale]

Or even also using audio labeling 😄

[pastel oak]

Yeah, audio labelling on audacity for example is a beauty

[odd shale]

If the pitch varies on the dataset or where?

[pastel oak]

Well if the audio you put in varies in pitches then youll hear it just like you hear it in the inferenced vocals with markiplier

O wait you meant dataset

ignore

[brisk quarry]

nop

i meant when the audio pitch varies

[pastel oak]

RVC copies the pitches of the audio you put in

so it will vary too

[odd shale]

Do you mean the dataset or the audio you use to test a model with?

[brisk quarry]

audio

so if i use more files with different pitches the model will also learn that?

[odd shale]

It should.

That would add more variety to the model.

[odd shale]

RVC will still try to mimic that audio's pitch when inferencing.

But depending on the results, you'll still must change the pitch/transpose.

[brisk quarry]

does rvc support .aac

[odd shale]

I'm not sure.

[knotty moth]

https://rentry.co/VoiceChangerGuide
https://rentry.co/forkvoicechangerguide
the latter is the optimized fork

[knotty moth]

you mean the pitch range? the more the better, and wide range of talking intonations and emotions can achieve it

[knotty moth]

converting to wav and using it as input is recommended

[brisk quarry]

why is it not making the index file after clicking some times

[torpid wasp]

How do i upload RVC model in hugging face?

[brisk quarry]

it dissappeared

there is nothing when i try to select it

[indigo creek]

how do i fix the lag?

are some ai model optimzed or nah

[pastel oak]

Are you talking about realtime voice changing

[indigo creek]

yh

the defult models work rlly good but when i install other people it lags

does it have to do with how people train the models or somethin

[pastel oak]

https://docs.aihub.wtf/essentials/voice-models/#uploading-to-hugging-face

[pastel oak]

I assume you have an AMD Gpu?

[indigo creek]

yh

[pastel oak]

AMD GPU users on original wokada v1 have to export the uploaded modells to onnx, and reupload those onnx models in the normal model upload slot

It would be better if you download the optimized wokada version, where you dont have to do this anymore

https://rentry.co/Forkvoicechangerguide

[indigo creek]

im so confused

idk what u just said

[pastel oak]

Epic

I explain

[indigo creek]

what did u say

[pastel oak]

Enjoy 🙂

[indigo creek]

do i have to install another program

[rare goblet]

Ayo? @indigo creek level 1 !!!

[pastel oak]

No, I gave you a recommendation because AMD GPU on the version of wokada you have is not optimized well

The other will work better for you

[indigo creek]

whatss wokada

[pastel oak]

...the voice changer

[indigo creek]

oh is it caleld wokada?

[pastel oak]

Yes

[indigo creek]

alr so after i uploaded my model

i need to click on export?

to onx

[pastel oak]

Yes. Then it creates an onnx file and downloads it to your Downloads folder. Then you go back to "edit" -> "upload file" -> you upload that onnx file

[indigo creek]

alr thanks

[pastel oak]

However, as I said, I recommend you download the other version of wokada where you dont have to do this, but up to you

[indigo creek]

what the other version of wokada

it's a bit confusing cuz there like 1000 downloads

of different versions

[pastel oak]

Nvm

[indigo creek]

alr thanks for the help shad tho

[brisk quarry]

why does my index file keep dissappearing

[ruby steppe]

@pastel oak how do you not make it so choppy i have a rtx 4070

[pastel oak]

Whats your F0 det, Chunk and Extra

And do you see your RTX 4070 in the GPU section or do you see buttons?

[ruby steppe]

yes i see it in the gpu section my extra is 4096 and my chunk is 320

[rare goblet]

Ayo? @ruby steppe level 1 !!!

[pastel oak]

You can do 96 chunk, 16384 extra and f0 det rmvpe

[ruby steppe]

alr

[pastel oak]

I assume you mean crackles with choppy, else tell me what you mean

This is the fix for crackles

[ruby steppe]

yes the crackles as well

what dose index do and whats a good set for it @pastel oak

[knotty moth]

you can also do the crackle fix trick by typing this in cmd/powershell (or make a bat file)
powershell "ForEach($PROCESS in GET-PROCESS audiodg) { $PROCESS.ProcessorAffinity=4; $PROCESS.PriorityClass='High' }"

[ruby steppe]

is there a way you can stop the cutting out?

[pastel oak]

Index is optional. This forces the accent of the voice model. It can help to make it more realistic or if you have a heavy accent and you want to reduce it (can cause artefacts if your pronounciation is vastly different)

I personally dont use it, but you may use something like 0.3 - 0.5

[pastel oak]

Your 4070 should be able to run the settings I told you to use - else maybe youor microphone volume is too low, and it cuts out? Did you move S.Threshold all the way to the right? If yes, move it back left

[knotty moth]

lower the threshold as needed

[pastel oak]

Oh thats cool thank you for the info

[ruby steppe]

ok its at 0.001 right now

so make it zero or js lower it

[brisk quarry]

everyone complaining

[pastel oak]

Move all the way to the left

I dont remember the metric system of wokada v1 so idk if those numbers are left or right lol

Dont think "zero" is possible

[ruby steppe]

all good it sounds so much better any rec girl voices?

[knotty moth]

https://tenor.com/view/skill-issue-urge-resist-cat-gif-25536160

[brisk quarry]

[ruby steppe]

@pastel oak Another question so im in discord where i can hear myself, and I can hear these voices is that the voice cahnger or

[pastel oak]

I dont understand the question sounds like youre hearing voices in your head

lol

Can u phrase it differently again

[ruby steppe]

sure lol

[rare goblet]

Ayo? @ruby steppe level 2 !!!

[ruby steppe]

alr so when you know how you can make discord where you can hear yourself. when i have the voice changer on i hear these whispers n shi idk how to explain it.

[pastel oak]

Your mic is picking up every sound in your room, may it be your chair squeaking or u tapping your desk etc

You can enable sup2 to help against it

[ruby steppe]

ohh ight thanks

[fallow parcel]

does w okada support amd gpu ? Cuz i can only select cpu

[pastel oak]

You downloaded the wrong version of wokada in that case.

For AMD GPUs we recommend the optimized wokada version which runs great. This runs on a webui instead of a client app, hope u dont mind. This helps with performance:)

https://rentry.co/Forkvoicechangerguide

Go to Downloads and get AMD one

If you insist on the client version, you can follow this guide. AMD has some little things to look out for in this one, like exporting models to onnx and reuploading them before being usable without lags, keep that in mind
https://rentry.co/voicechangerguide

[elfin quiver]

can anyone help me to make any sound to model for use in RVC-GUI?

[fallow parcel]

Works fine now thank you !

[brisk quarry]

amogus

[cedar imp]

Python path configuration:
PYTHONHOME = (not set)
PYTHONPATH = (not set)
program name = 'env\python.exe'
isolated = 0
environment = 1
user site = 1
import site = 1
sys._base_executable = 'C:\Users\me\Desktop\Applio-3.2.4\env\python.exe'
sys.base_prefix = ''
sys.base_exec_prefix = ''
sys.platlibdir = 'lib'
sys.executable = 'C:\Users\me\Desktop\Applio-3.2.4\env\python.exe'
sys.prefix = ''
sys.exec_prefix = ''
sys.path = [
'C:\Users\me\Desktop\Applio-3.2.4\env\python39.zip',
'.\DLLs',
'.\lib',
'C:\Users\me\Desktop\Applio-3.2.4\env',
]
Fatal Python error: init_fs_encoding: failed to get the Python codec of the filesystem encoding
Python runtime state: core initialized
ModuleNotFoundError: No module named 'encodings'

Current thread 0x00002990 (most recent call first):
<no Python frame>

Press any key to continue . . .

[brisk quarry]

Ok

[errant stump]

help [Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...

[echo coral]

@pastel oak dms

[crystal merlin]

Guys

How do I make the program's audio better?*

[brisk quarry]

Better samples

[crystal merlin]

Where?

[echo coral]

Anyone wanna help me get okada voice changer

[brisk quarry]

Make yourself

[brisk quarry]

No

Just download it

[echo coral]

I did but how do I get it working on discord

[brisk quarry]

Its not that hard

Idk

I dont use okada

[echo coral]

Oh which one do u use

[brisk quarry]

None

I hide myself from people instead of showing myself

[crystal merlin]

But isn't there an option that makes the program better, something like that?

[rare goblet]

Ayo? @crystal merlin level 1 !!!

[brisk quarry]

I dont think so

Its the model

[crystal merlin]

oh ok

[ivory ravine]

how do i start making my own voice models just like how they post it in? #1175430844685484042 ? any video guide on how to start will be great 🙏

[brisk quarry]

Idk

[ivory ravine]

👍

[odd shale]

You can start by reading the docs buddy.

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[ivory ravine]

much appreciated :D

[odd shale]

There are some W-Okada docs you can read buddy.

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[echo coral]

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[rare goblet]

Ayo? @echo coral level 1 !!!

[echo coral]

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[odd shale]

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[azure marsh]

Suggestion for @whole horizon

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[marsh nebula]

Crazy

[odd shale]

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[brisk quarry]

" You can fix that by using rx spectral de-noise, rx 11's dialogue isolate, mel/ bs roformer, bandit plus v2, uvr de-noise or clarity vx pro.*
"

What are those things

[crude flame]

those are all de-noisers or vocal separators

[thin stump]

Check your dms and i'll hook you up with nano-rizz

@brisk quarry hello?

[brisk quarry]

Hi?

[lapis bridge]

Why my rvc so quiet even with max gain

[coral sluice]

anyone know how to use an ai model thats been downloaded i just text to speech with it or wav

[tame mica]

what

[plain fulcrum]

Yall think a 30 min dataset with 500 epochs is enough for a really good model?

[odd shale]

Actually, dataset length doesn't matter that much but yeah, 30 mins is good.

But keep in mind the dataset's quality/how clean it is play a role on the model's final result.

[plain fulcrum]

Oh damn

I didn't know tht

[odd shale]

Do you mean W-Okada?

[plain fulcrum]

u think this is good enough quality?

i dont rly know the standard lol

[odd shale]

Yep, it's good enough but it seems your dataset is pretty inconsistent on quality/waveform.

[plain fulcrum]

oh yeah u right i gatta even them out a bit

quality wise i cant rly do much else since im pulling from different sources

[odd shale]

Personally i would suggest you to just use one voiceline pack.

[plain fulcrum]

so the dataset being inconsistent is tht bad of an issue?

[odd shale]

In this case, i think so.

But you can experiment anyway

For example, just use 2 voiceline packs from him

[plain fulcrum]

yeah i might remove the one in the beggining its very different from the rest of the packs i used

[odd shale]

Alright mate.

[plain fulcrum]

yeah thx for the help man i appreciate it🙏

[odd shale]

You're welcome bud.

[stray wind]

Hi, I'm on Mainline Kaggle and I'm on Process Data step and and you get "failed to load audio: ffmpeg error"

start preprocess
['infer/modules/train/preprocess.py', 'datasets', '32000', '2', '/content/training/logs/mi-test', 'False', '3.0']
datasets/Pop5->Traceback (most recent call last):
File "/content/training/infer/lib/audio.py", line 37, in load_audio
ffmpeg.input(file, threads=0)
File "/content/training/.venv/lib/python3.10/site-packages/ffmpeg/_run.py", line 325, in run
raise Error('ffmpeg', out, err)
ffmpeg._run.Error: ffmpeg error (see stderr output for detail)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/content/training/infer/modules/train/preprocess.py", line 87, in pipeline
audio = load_audio(path, self.sr)
File "/content/training/infer/lib/audio.py", line 42, in load_audio
raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: ffmpeg error (see stderr output for detail)

datasets/place-audio-here.txt->Traceback (most recent call last):
File "/content/training/infer/lib/audio.py", line 37, in load_audio
ffmpeg.input(file, threads=0)
File "/content/training/.venv/lib/python3.10/site-packages/ffmpeg/_run.py", line 325, in run
raise Error('ffmpeg', out, err)
ffmpeg._run.Error: ffmpeg error (see stderr output for detail)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/content/training/infer/modules/train/preprocess.py", line 87, in pipeline
audio = load_audio(path, self.sr)
File "/content/training/infer/lib/audio.py", line 42, in load_audio
raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: ffmpeg error (see stderr output for detail)

end preprocess

this is the message

can anyone help me?

[rare goblet]

Ayo? @stray wind level 1 !!!

[knotty moth]

you could do normalize those distinct four parts, also I sometimes heard some low-end rumbles/kicks

[stray wind]

have I placed my voice data file in the right place?

I don't understand what's the error

[tame mica]

no

you upload it to kaggle directly as a dataset in the menu on your right

[pastel oak]

The "pop5" is his dataset apparently. I dont know what audio file format that is, but he is unable to process data due to ffmpeg error

[stray wind]

I can't find this menu

where he's?

[stray wind]

it's in wav

[stray wind]

thank you I'll look later because I got disconnected from collab

[stray wind]

ohh I see thank you for your help

[tame mica]

oh for collab is different

upload it to your Drive instead

[knotty moth]

try re-export the wav file in Audacity, or if you can't open it, do export in another audio editor (izotope RX, adobe audition, etc) with the option "include metadata" or similar one disabled

[tame mica]

or rename it to add a .wav if it is already a wav lolz

[stray wind]

okok thank you

[stray wind]

okok I'll do it

[stray wind]

that's what I did

[rare goblet]

Ayo? @stray wind level 2 !!!

[night relic]

Hey,
Anyone a ESRGAN expert here??

[night relic]

Approach me through DM

[sick pawn]

Hey, I know how to make voice models and all that, but how can I use them directly for text to speech instead of just laying the model over a text to speech audio? Can someone help me out?

[rapid marten]

Hello. I'm kind of new to this AI real-time voice changer stuff, but I just really want to know. Is there any others realtime ai voice changer the support AMD? My PC is kinda old my GPU IS AMD R7 M440

INTEL HD GRAPHICS 620

I've been just losing my mind trying to find a good one.

Oh, and I know W okada exists, but I can't use it because I am on AMD and I am in Windows 11.

[pastel oak]

GPU is too weak, you can try the wokada-Fork which works great for AMD Gpu but i dont know how big the delay will be. You are still free to try. Else, if your internet is good (at least 100 mbps download) then you can try colab or kaggle.
Which one do you want me to link u

Actually, the gpu might be even too old for the wokada-fork 😅

Online is the only option I see for you

[rapid marten]

God fucking dameit oh wellI guess I'll take the online option But is it like a real-time voice changer or do I just put an audio and it converts it?

[pastel oak]

You can do both actually

But to answer your question you can use your mic yes

[rapid marten]

Ok good tell me what is it?

[pastel oak]

Colab (4 hours of daily usage)
https://colab.research.google.com/github/deiteris/voice-changer/blob/master-custom/Colab_RealtimeVoiceChanger.ipynb
You kinda just run all the cels one by one, you do need an ngrok account which is free so thats extra steps but also fine

Kaggle (30 hours per week, more complicated to set up but once you have it its easy)
https://www.kaggle.com/code/suneku/voice-changer-public
There is not a realtime-related guide for kaggle, but theres a guide for something else which instructions you can follow, but just use the kaggle link I linked you instead of cloning the notebook they have in the guide. Principle is the same
https://rentry.co/RVC-Mainline-Kaggle

[rare goblet]

Ayo? @rapid marten level 1 !!!

[pastel oak]

?

[rapid marten]

Ok ok I well try this

[brisk quarry]

get more data

[sick pawn]

Hey, I know how to make voice models and all that, but how can I use them directly for text to speech instead of just laying the model over a text to speech audio? Can someone help me out?

[brisk quarry]

data of obvious accent

[brisk quarry]

TTS tab....

do more then

or less

idk

maybe overtrained

i have no idea what rvc disconnected means

disconnected from what

make a new model with only liike 200 epochs and batch size 4

idk

maybe try like 10 minutes

now make me a sandwich

[thin stump]

The inference needs to be an accent. Say you have Tara Strong doing a cuban accent but if you use American inference then that's just Tara Strong https://www.weights.gg/models/cluyncyqj00e5w2lfxi7jyt25

[brisk quarry]

YO

METAL GEAR

i only played mgr tho

[thin stump]

☮️

[brisk quarry]

but when mgs4 release for pc i will play

im gay for raiden

mgs2 raiden looks like a femboy

[thin stump]

lol, his game is still very much peak. I liked him in 2 too but there's still this hate train

2 was always the best compared to 3

[brisk quarry]

https://tenor.com/view/raiden-msg-raiden-raiden-smoking-smoking-mgs-gif-26226025

[sick pawn]

Hey, I know how to make voice models and all that, but how can I use them directly for text to speech instead of just laying the model over a text to speech audio? Can someone help me out?

[thin stump]

batch 16 is less versatile than batch 8, but the accent is accurate. If you needed to use it for realtime, then always use batch 8 with minimum length datasets

Yeah it's fine. 4 will have shaky graphs I believe but it's 10 minutes of clean audio you said

I just recommend 8

always monitor g/total for the best lowest point so not 200 epochs

And you can use kaggle instead if rvc disconnected kicks you out (30 hours of GPU weekly) https://rentry.co/RVC-Mainline-Kaggle

[rare goblet]

Ayo? @brittle wing level 7 !!!

[brisk quarry]

TTS.

There is a own tab for it

[brisk quarry]

4 is more accurate tho

Id say 4 for accent nd 8 just for voice

I personally use 16

[sick pawn]

kein help channel falls es dir nicht aufgefallen ist

[brisk quarry]

?

Was laberst du

[sick pawn]

ist doch gut luka

erstmal brille putzen

[brisk quarry]

This guy is high

[sick pawn]

sure

[brisk quarry]

L

[sick pawn]

rip no gifs

for u luka

[brisk quarry]

https://tenor.com/view/saul-saul-goo-saul-goodman-meme-based-meme-gif-25971867

[sick pawn]

L for Luka

[brisk quarry]

Ok

[sick pawn]

🤓

[brisk quarry]

@red kayak can you ban this troll

[thin stump]

Your only option is elevenlabs

[sick pawn]

"this troll" is the owner of ai hub germany

[sick pawn]

thanks

[brisk quarry]

💯

[red kayak]

whats wrong?

[brisk quarry]

That server would suck

[thin stump]

Nep is just a troll himself lmao please ignore

[brisk quarry]

Read it all yourself and translate the stuff

What

[brisk quarry]

Ur also high

Why is everyone high today

[sick pawn]

"crash 4 grinding time trials"

🤣

2006

[brisk quarry]

@red kayak see

[red kayak]

i dont get what they are trying to convey

[brisk quarry]

They are making fun of like everything in my profile

[sick pawn]

pov you are nep: 😔 😢 😟 😤 🤬

[brisk quarry]

[red kayak]

alright settle down for now

if u 2 want to argue, u are free to do so in dms

this is a help channel

[sick pawn]

you arent even on the private ai hub server why

[brisk quarry]

I dont want to talk to him

[red kayak]

keep this up and i'll have to time out te both of u

[brisk quarry]

WHAT DID I EVEN DO

[sick pawn]

keep crying wtf just block me

[red kayak]

dont think we are affiliated...

[sick pawn]

yes we are

[brisk quarry]

bet

[red kayak]

well i said if u keep pushing the issue

[brisk quarry]

I said ban him to not push it

He will continue

[red kayak]

same goes to him

[sick pawn]

ich weiß das du es bereust

[sick pawn]

i didnt do sh*t tho

[brisk quarry]

[red kayak]

you are arguing thats enough of a reason for me to mute

[sick pawn]

clicked on blocked message before taking the screenshot

[red kayak]

dont cause drama

[red kayak]

stop

literally, if you want to beef with him go into his dms

not here please

[sick pawn]

no worries i wont write stuff to people like him, hes just beeing weird and doesnt help. ill stop

[brisk quarry]

I blocked him

[thin stump]

Anyways, as I was saying elevenlabs. If there was a better TTS alternative we would know about it

[red kayak]

okay thank you, he also blocked you so then again it shouldnt be an issue

[red kayak]

gpt sovits if u want something 4 free

or fish

[brisk quarry]

@thin stump applio haa tts built in

[sick pawn]

i thought there was a thing that directly used ur model

so i need to use elevenlabs and convert with my model?

[red kayak]

thats tts audio and then infered with an rvc model

[brisk quarry]

Isnt that was hes asking for

[red kayak]

not legit tts off the voices u train which is a little lame

[brisk quarry]

Hmm

[red kayak]

ahh

[red kayak]

yeah u are right

[red kayak]

he wants that

[brisk quarry]

Im always right

(False)

Im watching star strek rn

[red kayak]

you could either train an 11labs model or take one of their models, infer with it and then just use that as a tts by applying over ur voice model

[sick pawn]

i have like 50 selftrained high quality rvc models, i wont train them new 😭

[brisk quarry]

Skill iss-

[sick pawn]

see

[brisk quarry]

Whateven is that

[sick pawn]

clown

[brisk quarry]

Clown rat

[red kayak]

high quality is objective but what i will tell u is that you can just use those just fine by infering audio from elevenlabs and then running ur models through that audio

Literally stop acting like toddlers

[brisk quarry]

My last words

[sick pawn]

i stopped

[red kayak]

I've asked u both 3 times

[red kayak]

just block him

[red kayak]

better be

[sick pawn]

i did like 1 min ago

[red kayak]

anyhow yes this ^^

[red kayak]

we love drake yes

[sick pawn]

ill try that, even tho i already did it like that back then

[brisk quarry]

Thats drake???

He looks kinda black

[red kayak]

well yeah, but if u want. You can pay and trian 11lab models instead thugh

[brisk quarry]

Did drake turn black

[red kayak]

those will be more robust

[red kayak]

go to https://discord.com/channels/1159260121998827560/1159265300915687534 you goof

[brisk quarry]

Lmao

Me goof

[red kayak]

yes uwu

[brisk quarry]

@sly yew

[toxic coral]

when i try to open start_http or start_http_bat it cloes soon as i open it

[cosmic epoch]

How do i train a model in this hugginface? https://huggingface.co/spaces/r3gm/RVC_HFv2

how do i upload dataset to here?

https://huggingface.co/spaces/r3gm/RVC_HFv2

where do i need to click in order to add dataset or audio files to train in this hugginface? https://huggingface.co/spaces/r3gm/RVC_HFv2

[odd shale]

Can you calm down?

[cosmic epoch]

sorry

[rare goblet]

Ayo? @cosmic epoch level 2 !!!

[odd shale]

And no, you can't train models on Huggingface.

[cosmic epoch]

but there is an option of traning in this one. look at the link

[odd shale]

That doesn't mean you're able to train on HF.

[cosmic epoch]

so, are there any colabs where i can train models? besides the applio one?

[odd shale]

Yep, there are, including Kaggle notebooks.

Take a look at the guides.

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[cosmic epoch]

can you please give me a link to the Kaggle notebooks?

[odd shale]

-kaggle

[azure marsh]

Suggestions for @cosmic epoch

📘 Kaggle Notebooks

• 🆕 Applio Notebook by Vidal
• 🆕 Applio Notebook by Shirou
• 🆕 Audio Separation by Shirou
• UVR5 NO UI by Eddy
• How to use RVC Mainline Kaggle by Cauthess
• ✨ RVC Mainline by Hina
• Original W-Okada's Voice Changer Kaggle
• Modified W-Okada's Voice Changer Kaggle

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

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

[cosmic epoch]

i did all the steps in the Mainline_Colab, but when i run the last cell, it only shows me the local gradio URL, which i can't acess. how can i acess the MAIN gradio URL where i can train voices?

[odd shale]

Did you make sure to put your ngrok token on the last cell?

Do you have an ngrok account?

[cosmic epoch]

i do.

should i write Ngrok_Token in the name of the notebook acess? because i did that already

[odd shale]

In between those spaces, maybe you didn't put your ngrok token

[thin stump]

use the kaggle instead if you're already going through the colab guide

its the same steps

[cosmic epoch]

ohhhhh, i get it now.

[odd shale]

Are you using kaggle or colab?

[rare goblet]

Ayo? @odd shale level 80 !!!

[cosmic epoch]

colab. why?

[odd shale]

I thought you were using kaggle but well.

It's the same steps.

[low shard]

kaggle offers better performance yea

[toxic coral]

when i try to start the voice changer dont work

why is that

[odd shale]

Where are you trying to use W-Okada?

Local, colab or kaggle?

Are you using the OG version or Deiteris fork?

[cosmic epoch]

okay, i now gotten into the gradio, but when i click on train model, it does the Traceback (most recent call last):
File "/content/training/.venv/lib/python3.10/site-packages/gradio/routes.py", line 437, in run_predict
output = await app.get_blocks().process_api(
File "/content/training/.venv/lib/python3.10/site-packages/gradio/blocks.py", line 1346, in process_api
result = await self.call_function(
File "/content/training/.venv/lib/python3.10/site-packages/gradio/blocks.py", line 1074, in call_function
prediction = await anyio.to_thread.run_sync(
File "/content/training/.venv/lib/python3.10/site-packages/anyio/to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "/content/training/.venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 2177, in run_sync_in_worker_thread
return await future
File "/content/training/.venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 859, in run
result = context.run(func, *args)
File "/content/training/runmain.py", line 488, in click_train
& set([name.split(".")[0] for name in os.listdir(f0_dir)]) error. How do i solve that?

[odd shale]

Did you even upload the dataset, preprocessed and feature extracted it first?

[cosmic epoch]

i did.

[odd shale]

Then i don't know why that error happens..

[novel river]

how can i use a custom model for ilaria rvc that has no index file?

it gives me an error when i have a pth file only

[pastel oak]

Upload a random index file, then set the index file usage to 0

[vale osprey]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer 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, -overtraining, -realtime, -rvc, -help

[fallen glacier]

Does anyone know how to fix the beat rice problem, where I can only use Beatrice?

[violet heron]

What’s your GPU?

[knotty moth]

beatrice trainer? no one has made it so far

[toxic coral]

somone help

[violet heron]

What do you need help with

[fallen glacier]

GPC, I use CPU

[fallen glacier]

No, beatrice models

[violet heron]

You probably can’t use it unless your cpu is AMD

[spring orchid]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer 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, -overtraining, -realtime, -rvc, -help

[knotty moth]

no, there's only built-in beatrice model, not custom ones made using beatrice trainer

[knotty moth]

I think they are looking for (custom?) beatrice models

[night relic]

Hey Everybody,
I have a google collab notebook which upscales images and videos using ESRGAN and RealESRGAN.
You have to mount google drive for giving the paths to the destination folder of the model and media(image/video) in the collab notebook cell.
On images it works excellent, while on the video tab, it shows the following error.
I would provide you with the collab notebook code and screenshot for analyzing the issue.
Kindly cooperate with this issue.
The first two images are of the collab and the 3rd image is of the video upscale issue.

https://paste.pythondiscord.com/WQKQ

Kindy copy paste the collab code and mount your drive and model in cell and run the cells.

[umbral anvil]

Anyone know how to use a gpt sovits model like this? https://discord.com/channels/1159260121998827560/1272353975886155848
I try to run it but theres no config file
Theres a .ckpt file? can I somehow use that insteado f the config file

[brittle wing]

If I use a 1.5-hour voice recording, will the AI generated voice sound 100% like mine? What data is needed to make the AI perfectly mimic my voice?

[thin stump]

-good mic quality
-try to record in a padded room
-at least 10 minutes, not monotone. Try to put some emotions into it if you want when you do some line reading

So yeah, the more noisy the environment = more clean up needed on RX11

[bold ermine]

Guys any update on API for RVC

[brittle wing]

yo, whats the best model on mvsep for removing bg vocals?

[night relic]

Hey Everybody,
I have a google collab notebook which upscales images and videos using ESRGAN and RealESRGAN.
You have to mount google drive for giving the paths to the destination folder of the model and media(image/video) in the collab notebook cell.
On images it works excellent, while on the video tab, it shows the following error.
I would provide you with the collab notebook code and screenshot for analyzing the issue.
Kindly cooperate with this issue.
The first two images are of the collab and the 3rd image is of the video upscale issue.

https://paste.pythondiscord.com/WQKQ

Kindy copy paste the collab code and mount your drive and model in cell and run the cells.

[surreal pine]

start preprocess
['infer/modules/train/preprocess.py', 'dataset', '40000', '1', '/content/Ilaria-RVC-Mainline/logs/🫳', 'False', '3.0']
dataset/.ipynb_checkpoints->Traceback (most recent call last):
File "/content/Ilaria-RVC-Mainline/infer/lib/audio.py", line 61, in load_audio
with open(file, "rb") as f:
IsADirectoryError: [Errno 21] Is a directory: 'dataset/.ipynb_checkpoints'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/content/Ilaria-RVC-Mainline/infer/modules/train/preprocess.py", line 87, in pipeline
audio = load_audio(path, self.sr)
File "/content/Ilaria-RVC-Mainline/infer/lib/audio.py", line 73, in load_audio
raise RuntimeError(traceback.format_exc())
RuntimeError: Traceback (most recent call last):
File "/content/Ilaria-RVC-Mainline/infer/lib/audio.py", line 61, in load_audio
with open(file, "rb") as f:
IsADirectoryError: [Errno 21] Is a directory: 'dataset/.ipynb_checkpoints'

dataset/🫳.wav->Suc.
end preprocess

[proven hill]

try not having an emoji as the name of the dataset

[surreal pine]

i didn't, i just replaced the name with an emoji because i don't really want to show it

it doesn't have any special characters except for - or spaces

[proven hill]

is it on colab?

[surreal pine]

yeah

[proven hill]

try deleting the runtime and restart

[surreal pine]

alr

[naive fossil]

what does it work atm with paperspace?

i dont come in touch with rvc or applio since Feb-Mar

[polar pier]

ใ

[stray wind]

I tried again but it still doesn't work

[tame mica]

no put it in your drive

google drive

[stray wind]

oh and where in gg drive

[tame mica]

anywhere

[stray wind]

it's in my drive

but I don't understand why it doesn't work.

did I make a mistake?

[odd shale]

Yes you did.

You had to write the path as datasets/DataPop

[stray wind]

yes and what's the good path

?

[odd shale]

The path you must write is datasets/DataPop as i mentioned.

[stray wind]

ohh okok thakn you for your help

(I really don't know much about computers)

I got this

Have I made another mistake?

I really don't understand why it doesn't work

[odd shale]

No, this means it's working.

The time it takes to preprocess will depend on the dataset's lenght.

[stray wind]

ohh ok so I can do the step 2b?

[odd shale]

Wait till the output ends preprocessing

[stray wind]

okok and you know how long it might take?

I've 15min of dataset

it's all over?

[pastel oak]

yes worked

[stray wind]

ok thank you

I just clicked on “Train model” and got this

[rare goblet]

Ayo? @stray wind level 3 !!!

[stray wind]

do we agree that my model is being trained?

[pastel oak]

Model is training

[stray wind]

ok thank you

an do you know where I can download the files of my model?

it's this file the ".pth file"

?

where can I find the ".index file"

?

I don't find it

[tame mica]

it's not finished yet lolz

[stray wind]

oh then I can't download the index file

because when I was training on rvcDisconnected I can download it

[tame mica]

did you click train index or train first?

[stray wind]

"train model"

nahhh my collab crash, does anyone know why collab crashed, I was told that this collab didn't crash unlike rvcDisconnected?

[indigo hinge]

hello i have 2 file index e pth of emila voice from re:zero. But I don't know what to do. I can get some help ?

[tawdry meteor]

I'm just starting out, could anyone guide me on how to use the model trained in RVC, converting the Onnx model to Pth?

[violet heron]

Do you want to use the voice in realtime or for covers

[thick dagger]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[rare goblet]

Ayo? @thick dagger level 3 !!!

[thick dagger]

-gui

[azure marsh]

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

[thick dagger]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer 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, -overtraining, -realtime, -rvc, -help

[final sail]

How to install mangio fork on linux

[finite prawn]

where does applio put index files when generating index? i just finished a model but i cant find the index

[granite bobcat]

hello i'm new and i have no idea how to use the ai model i just downloaded can someone help?

[odd shale]

Actually you just need the model's huggingface link if you would like to use it online.