✨│ai-help

[rare goblet]

Ayo? @lone iron level 2 !!!

[glad zealot]

they didnt even change anything

bru

[low shard]

exactly it doesnt even work for phones lol

ye fake asf lol

Dont follow yt tuts

Unfortunately, there is no possible way to use realtime voice changer on phone, no matter what

[glad zealot]

even my gura notification is there XD

like literally nothing was even removed

[low shard]

bomb comment his video lol

[lone iron]

Hopefully i join this server can solve pyngrok not found problem :)

[glad zealot]

meh it will break eventually

[lone iron]

real?

can you send the newest link one? for pc

[low shard]

yea, the one used by the video is just a copy of @glad zealot colab

this one works

[glad zealot]

ya, i mean you could technically use it but only for recording, for actual calls and real time it wont work

[lone iron]

So this only work on pc?

Okeey man thanks you guys for helping me

[low shard]

yep

only pc

there is also a kaggle version that gives u more time of gpu for free (colab gives 4 hours daily, kaggle gives 30 hours weekly) but its harder, if u want i can send that too

[lone iron]

Nah if that more harder i pass 😭

I will use that one

[low shard]

alright, anyways u can bypass the 4 hours daily by making google alt acc if u hit the limit, also the guide for it is https://docs.google.com/document/d/e/2PACX-1vTIceEcBfS6Zqolv_QEysrFfI_EJikPxozWptP_EjkpLVl-l-gdo-ijBonQMTviAHEYm5emmd9k9TdC/pub

[lone iron]

So thats mean the youtube source is the modified one?

[low shard]

its not even modified, its an old copy that doesnt even work lol

[lone iron]

Oh thats why pyngrok not found?

[low shard]

yep

he literally used old code, and just changed the name of the colab

total scam for just views

[lone iron]

I see" this is the newest one right?

[low shard]

yep, i also sent u a written guide as it could help you to use the colab

[lone iron]

Aight thanks you so much man for helping me out 😭 ❤️

[low shard]

your welcome

[lone iron]

Btw last one question should i check mark the play notification?

[low shard]

all it does is display a sound notification to remind u its done running, its really ur choice

[lone iron]

Aight thank you once again bro

[low shard]

yw

[glad zealot]

@low shard just noticed its already broken so i didnt have to do anything XD

[glad zealot]

dont work properly XD

[low shard]

u mean the copy? ye its broken lol

bro couldn't even copy properly

[glad zealot]

ye

the notif is not working properly too since if the page is hidden/not focused when its done

it wont play

doesnt really matter since the voice changer will still just, around 5 mins is teh boot time

[low shard]

i tried commenting its fake and my comment got filtered out 💀

[lone iron]

In where? 💀

[glad zealot]

lmao

[low shard]

in the yt vid comments, i tried commenting it, refresh and see by latest and couldn't see mine

[glad zealot]

use a different font :>

[lone iron]

In this one?

Question if thats work on phone it is take a lot storage?

[glad zealot]

no

it wont use any storage

[low shard]

worked lol

[low shard]

nope, it's on cloud (meaning that the program runs on a remote good pc, not your pc)

[lone iron]

No im asking if i tried on phone

Because my friend asking me how to do use that rvc okada on phone

[low shard]

if u tried on phone u couldn't use it for realtime, and yea still its on cloud so wont take any storage

[lone iron]

For only record

[low shard]

for only record yea can work, wont use much storage

even if tbh for pre recorded audios i'd suggest other things personally

[lone iron]

Aight aight thank you once again sorry if i am asking to much

[rare goblet]

Ayo? @lone iron level 3 !!!

[low shard]

nah dw its fine

[lone iron]

So its still work right for only record tutorial for my friend

[glad zealot]

aight bed time for me

exam in 8 hours

weeeeeeee

[low shard]

rip

[low shard]

the updated colab can work for only record on phone yes

[lone iron]

Whats the setting 💀

[low shard]

Sorry i dunno about using it for pre recorded audios 😭 i remember its possible but i dunno personally sorry

[lone iron]

Nvm in that youtube vid there is the setting

Its still work for the setting right?

[opaque geode]

Hello! Can anyone tell me the difference between smaller epoch and larger epoch?

[odd shale]

What do you mean?

An epoch doesn't always dictate a model's quality.

The quality will mostly depend on the dataset you use.

[opaque geode]

What’s the difference between training a model to 300 epoch and 500 epoch?

[low shard]

epochs don't mean quality, its a unit of measurement of the ai training cycle, its better u see https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/

[low shard]

should

[odd shale]

Yep, what Nicky said.

[low shard]

he manually deleted them after 5 mins 💀
EDIT: nvm one stayed

[opaque geode]

Thank you

[rare goblet]

Ayo? @opaque geode level 1 !!!

[opaque geode]

I trained a few mobile legends model (which I cannot find anywhere) and some of them are quite dissatisfying

[odd shale]

Probably the datasets you made weren't properly clean or had weird effects on them.

[opaque geode]

I see.. I only have like 2 mins of dataset

[odd shale]

The short length of the dataset could have been other reason.

[lone iron]

Why i cant heard my voice in phone?

[opaque geode]

Because the character only have 2 mins of lines

[odd shale]

Are you trying to use W-Okada on phone?

[lone iron]

yes

my friend asking me how to use that on phone

but my server io didnt load my voice

[odd shale]

W-Okada isn't meant to be used on phone.

[lone iron]

This one they both said can use on phone for tutorial

And i want giving tutorial to my friend

[odd shale]

That's a blatant lie.

That tutorial is fake.

[lone iron]

No its from moderator lol

[odd shale]

?

In few words, no

You can't use W-okada on phone

[lone iron]

Read again he said can on phone for tutorial

[odd shale]

I see.

Well, you can actually use RVC on phone.

But you can't use W-Okada.

[lone iron]

But my serverio doesnt load my voice

I use rvc from him

[odd shale]

retrieval voice conversion, right?

Not voice changer?

[lone iron]

Do you know how to fix the problem serverio doesnt load it?

[odd shale]

I don't know..

[lone iron]

Aah how i giving tutorial to my friend

[odd shale]

Sorry..

[lone iron]

Its okey tho

Why now its said err ngrok 3200

[low shard]

for pre recorded audios

its better to use other things instead

use ilaria rvc zero which is even faster

[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! 💖

[lone iron]

Can use for tutorial on phone right?

[low shard]

yes it works on phone all fine, this is a fork of rvc that runs on a faster gpu

its different that wokada that is made specifically for realtime, wokada uses RVC models, as rvc is the main program

[lone iron]

wait bro why my serverio doesnt load my voice?

like for test the character ai voice that i choice?

[low shard]

wdym?

Are you using the wokada colab on ur pc and the voice is not working on discord ?

[lone iron]

For tutorial on phone

I already record it but its doesnt load the character ai that i choice

[low shard]

its better to not use wokada colab on phone

[low shard]

try this one instead for pre recorded audios

[lone iron]

How to use it

[low shard]

there's the guide in the message

[lone iron]

More simple?

[low shard]

there isn't any other guide than that, its better to try to re read the guide in the message that u can get by clicking the blue 'CLICK HERE'

[lone iron]

The link that you gave me it is wokada colab?

the first one

[low shard]

the guide of before is wokada colab, that u can use on pc yes

[low shard]

but for pre recorded audios, its better u use this one instead

[lone iron]

But how they do the tutorial on youtube using that wokada colab on phone?

[low shard]

its fake

it works only for pre recorded audios and not worth it, but its impossible to use wokada realtime on phone

as i said, there is no possible way to use realtime for games/call on phone, u need atleast a pc

[lone iron]

Watch the vid bro in last vid he can use the character ai that they choice

[low shard]

as you said your friend needed it for pre recorded audio, tell him to use this

[lone iron]

But when i try it i cant use or doesnt load the character ai

[polar crater]

Hello, I made a sound model from a video (I didn't post the video link because I didn't know if it was forbidden to post it). The sound model was downloaded as tar.gz, but how do I upload it to the RVC GUI? I'm sorry my english is bad

[low shard]

It cant work on phone

its a fake video just to get views

[lone iron]

So only for pc?

[low shard]

yes

ONLY PC

[lone iron]

Okeey okeey got it

thanks again

[low shard]

If you want to use the voice models for pre-recorded audios, use ilaria rvc zero instead

[low shard]

yw

[low shard]

tar.gz ? How did u train that model? And also be sure to not use RVC-GUI as its outdated

[polar crater]

I made with google colab

[low shard]

send the link

99% of youtube vids are outdated, never follow them

[polar crater]

https://www.youtube.com/watch?v=QgN7QgUJObE

in this video

[low shard]

dont follow yt tuts

You don't got a good pc right? as google colab is a cloud service for people who dont got one

[polar crater]

no ı have good pc

put how can ı do

[low shard]

whats ur pc gpu?

If you got a good one, you can do it locally (on ur pc), else u will have to use cloud services (using remote good pc)

but i need to know the name

[polar crater]

3,50 GHz

[rare goblet]

Ayo? @polar crater level 1 !!!

[polar crater]

how can I do

[low shard]

what

I need the GPU name, not the cpu clock speed

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

[polar crater]

6 core

CPU

AMD Ryzen 5 5600 6-Core Processor 

Base speed: 3.50GHz
Slots: 1
Cores: 6
Logical processors: 12
Virtualization: Disabled
Hyper-V support: Yes
L1 cache: 384 KB
L2 cache: 3.0 MB
L3 cache: 32.0 MB

Usage 5%
Speed ​​3.96GHz
Working time 0:02:50:00
Transactions 278
Thread 5585
Identifiers 150283

[low shard]

that's a cpu not gpu

[low shard]

its better u send a screenshot

[grim bay]

How can I download UVR-BVE model in UVR??

nvm, it's downloaded

[polar crater]

ı have to go ı will come back

[low shard]

alr

[lone iron]

How long its take to convert the voice? its been 300 s still doesnt convert yet using this one https://huggingface.co/spaces/TheStinger/Ilaria_RVC

[rare goblet]

Ayo? @lone iron level 4 !!!

[lone iron]

Its only 4 sec voice tho why its take to long

[low shard]

did u duplicate the space?

don't do that, you have to do it on the normal link

[narrow sentinel]

thx

[gloomy holly]

why don my model work? https://huggingface.co/greplol/mikey/resolve/main/mike.zip

[night rune]

-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, 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

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

[fickle sky]

can someone please send me the client download for NVIDIA GPU? i can seem to find it

[lone iron]

Bro why it is keep doesnt load or play my voice and convert voice?

[dreamy nymph]

I have an issue with a model, when I use it the voice just cuts as if some sort of envelope is acting

[rare goblet]

Ayo? @dreamy nymph level 1 !!!

[dreamy nymph]

it doesn't happen with other moels I trained

[dreamy nymph]

In which part do I apply that?

[knotty moth]

mb doesn't seem from applio split audio, but that sounds more like the input audio is overly gated, or prob the model dataset itself?

[dreamy nymph]

hmmm I see

thats quite deceving taking in Mind how long It took me to train It

[knotty moth]

using Renegate is recommended, and beware of silent voices that might fall below -40 dB, or simply normalize the audio
https://rentry.co/RVC-dataset-RX11#noise-gating-and-audio-labeling

[dreamy nymph]

Thanks, Maybe if I also train the index It might help it

[honest anvil]

hey I need some help with the program... its working properly but when it comes to acually using the mic in discord of whatever its just my normal voice instead

[knotty moth]

• the passthru button in the voice changer should be green, not red, click it if not yet
• double check the audio input & output (virtual cable) in the voice changer and discord

[honest anvil]

i need to download the virtual cable then ?

[knotty moth]

virtual cable is needed to route the voice changer output to discord or any apps/games
https://rentry.co/VoiceChangerGuide#virtual-audio-cables-mandatory

[wild bobcat]

does rvc have like a vowel system similiar to utau

[low shard]

could u be more specific ?

[tame mica]

like a, i, u, e, o?

[wild bobcat]

that and the

yknow like 10 other possible vowels like the E in dead and the A in and, etc

[tame mica]

it learns phonemes from the dataset

if u have those it'll be fine

[wild bobcat]

its just a question i had

im curious about learning about rvc and its quirks however

i cant find indepth stuff about it

and dont have many people to talk to about it

[wild bobcat]

and the finished ai voice tries to match the phonemes of each sample its given right

not every voice says vowels the same way

[tame mica]

yee

[wild bobcat]

i wonder

what vowels or voiced consonants can ai voices not do good with

like i think know that "N" and "M" are kinda difficult for the voice sometimes because theyre very similiar

[crude bolt]

Try different pre-trains

But you need to take everything into account - not just the trained model

This includes also the way you pronounce things. And the quality of the sound your microphone produces

I experienced better results with training a model based on the voice of a person from my country - so that all my false English accents are understood by the voicechanger

[knotty moth]

seems more like lack of dataset variation and voice clarity, though the default pretrain usually works fine

[weak meadow]

Hi, can someone help me?

[crude bolt]

Depends on your question

[knotty moth]

!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.

[weak meadow]

do we know how to configure it? idk its a voice.

Info: Batch Size: 8 Dataset Length: 30 Minutes Hop Length: 32 Pretrain: DMR Precision: FP32 Sample Rate: 32K

[low shard]

that's the information of how it was trained, if you are just going to use the model u don't have to configure anything

[weak meadow]

Ok, but there is no index?

[rare goblet]

Ayo? @weak meadow level 1 !!!

[low shard]

Where did u find the voice? could u send a link?

If there's no index and only the .pth u can't really do much about it, its a thing only the creator can train

[weak meadow]

I found the voice on the voice model e-girl chatroom

[low shard]

what

[tired ore]

Can someone help me? I installed everything correctly, both apps, but my voice doesn't come out normally or in DC. I activated the Cable but it doesn't work.

[brittle wing]

alr imma be real here I just decided to try some stuff out and I don't have the beefiest pc at all and I'm using replay's model maker which prob isn't the best and all but I'm not home often so I just have it running in the background. I've a 6gb 1060 and I've been running on around 48m of data, and it's been going on for almost 12 hours at this point, and it's at epoch 18. any recomms would be nice and all, if I should change from replay to something else. thanks for taking your time to respond, if anyone responds

[trim sparrow]

my google collab (applio) have disconected while I trained a voice do you know if it's possible to restart the training where its stoped or if I have to re train from 0 (I am on public link)

[crude bolt]

Since you voice don’t come out normally or in discord, verify first if your mic is plugged in

[proper shale]

Huge amounts of data on an older card like that are very slow, yes. I think reducing dataset size should be the best course of action

[proper shale]

If it saves automatically to Drive... yes, you can retrain it

[crude bolt]

Are you sure you are training with the gpu? By default the cpu is selected

(In that case yes it takes quite long per epoch)

[knotty moth]

5-10 mins of dataset should be enough given your weak gpu

[brittle wing]

CUDA was selected so that should've been the GPU and yeah, I just thought I'd go with 30m-1h but yeahh xD I'll try reducing my dataset and everything, thanks all for the advice, really appreciate it.

[knotty moth]

a cloud alternative is a kaggle notebook: https://rentry.co/RVC-Mainline-Kaggle
it has 30 hour weekly GPU quota (resets every Saturday 12am UTC+0), and it'd be good enough to train more than 30 min dataset

[dawn tendon]

Can someone help me? I pressed convert

But it loaded and said something went wrong

Time out UPLOADING data to server

What does it mean?

[carmine hearth]

#📰│dev-updates message

[dawn tendon]

Thanks

I just wasn't here for a while

[solid knoll]

Hey im new here, is there a guide on how to change my voice into a different character somewhere please?

[dull scaffold]

hey should i use 40k or 48k sample rate if the sample rate of my audio is 44.1k?

[lone iron]

Like i cant play my own voice and the convert ai voice

Btw how to make it more smooth voice? like the settings

[low shard]

ilaria rvc zero is not for realtime, it's only for pre-recorded audios

[lone iron]

Yea but i cant heard my own voice

[low shard]

You'd have to record your voice yourself, then upload the recorded file in ilaria rvc zero to convert it to the ai voice model

[lone iron]

Yea ik already but i cant heard my own voice after record my voice

Like the timer wont increase so literally i cant heard my own recorded voice

[knotty moth]

it may be either 40k or 32k depending on the spectrogram cutoff (open it in spek or audacity and see)

[lone iron]

How to make it more smooth voice its possible or nah? 🗿 whats the setting 🦖

[dull scaffold]

im guessing 32k then?

[ember ermine]

i searched messages and it seems the main fork of RVC is broken or something ? is there an alternative version other than Ilaria?

i couldnt get illaria to work

nvm, what i did was i downloaded their repo and tried running it locally

is there a local RVC that works?

[odd shale]

Yep, 32k

[odd shale]

Of course.

-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
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[ember ermine]

ty

[warm swan]

hey it's me again
Quick question:
Which Stem method do you guys recommend for the Replay ai
(feel free to ping on reply <3)

[little ore]

which model best for voide seprate?

[low shard]

It's better you don't use the UVR in Ilaria RVC Zero as it runs on CPU not GPU (so slow), use https://huggingface.co/spaces/TheStinger/UVR5_UI instead, and also u can read which are the best models here: https://docs.aihub.wtf/rvc/resources/vocal-isolation/#best-models

[knotty moth]

I'd recommend UVR with beta roformer patch or mvsep.com also the one just above (online)

[warm swan]

tankies

[fiery raptor]

can I ask about gpt sovits here?

[brittle wing]

-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, 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

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]

which colab is to train a voice?

[brittle wing]

@tiny mantle hi

[fiery raptor]

i was wondering because when i press on the go_webui.bat it doesnt open any web on page

[half bramble]

What's the link to the hugging face voice changer

[tame mica]

[rare goblet]

Ayo? @turbid tiger level 1 !!!

[indigo garden]

im having trouble installing tensorboard could i have some help?

[indigo garden]

训练结束, 您可查看控制台训练日志或实验文件夹下的train.log HELP

[odd shale]

There's no huggingface link to W-Okada

You either use it locally or via colab/kaggle.

[rugged oxide]

can someone tell me why the voice changer is saying trial

[rare goblet]

Ayo? @rugged oxide level 1 !!!

[rugged oxide]

like an input voice

an ai robotic girl is repeating the word trial

[proven hill]

wich one

[glad zealot]

You are using the free/trial version of vac by muzychenko from the sound of what's happening

Not sure how to get rid of it without using the paid version tho

[knotty moth]

mb, the VAC lite version from the link in the guide should fully work, no even need to get the Ahoy'd paid version

[meager comet]

is there any way to utilize npu on an ai cpu with rvc realtime

[rare goblet]

Ayo? @meager comet level 1 !!!

[dull scaffold]

is it possible to stop training on the mainline rvc and continue training after restarting your pc?

[knotty moth]

it will continue training from the latest checkpoint unless you lose the G and D files in the model's logs folder

[knotty moth]

none yet unfortunately in the fork wokada as well, not even ARM snapdragon cpus

[dull scaffold]

would i lose it if i just close the console? Because "Stop Training" doesnt appear in the web page for me

[rare goblet]

Ayo? @dull scaffold level 2 !!!

[dull scaffold]

hooray level 2

[narrow vortex]

how do you get the "model maker"?

[patent mauve]

yo what client do u guys for the rvc models?
im new help would be appreciated<3

[wild bobcat]

why do models sound so weird when they go under the c3 range? just asking out of curiousity

[brittle wing]

Is applio on pinokio also hacked?

[wild bobcat]

i dont know

[next geyser]

some voice models dont have index file in them

what should i do

[wild bobcat]

some are actually able to sing down to c2 but

they sound so glitchy and robotic as you can see in this audio

the first note is c3 the one after is c2

[wild bobcat]

i wish i could help you but i dont know

[next geyser]

dang

imma just get another index

[wild bobcat]

okay

[rare goblet]

Ayo? @wild bobcat level 7 !!!

[marsh python]

-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 Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline Kaggle by Cauthess

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

[next geyser]

\

how to use rvc disconnected

for a pth file

[plucky umbra]

what is "hop length"?

[golden walrus]

guys, i have issue where my output volume is way to small

is there a way to fix it ?

[fleet swallow]

trying to use mainline rvc on kaggle to do inference

not sure where to put my model file

[cinder wyvern]

can we still use the applio google colab while this situation is going on

[glacial pollen]

resolution / window / time frame of analysis window.
Analysis for grainy details, features and f0 / pitch variance or pitch jumps

Applicable only to crepe ( or mangio crepe if whatever you use utilize it that way )

tl;dr:
Smaller value = More accuracy but higher chance of " catching " impurities in audio
Higher value = Less accuracy but higher robustness

You almost never wanna go below 32 ( or even 64 ) and never above, I'd say.. 200 because if that's the case? just go for rmvpe which is anywhere from 200 to 500, don't remember.

32 <-> 156 is the range is I recommend where 64, 128 are the typical go-for values

[shadow atlas]

hi, im trying to get a realistic ai voice, it should be good for social media but i dont want any celebrity's voice. I want something like will from elevenlabs so what exactly do I do(i want to make it using my own python program)

[fleet swallow]

ok i found out where to put it but there's a new problem

i can't upload any files through imjoy elfinder

they all get stuck on "verifying upload file name"

[clear citrus]

Can someone give me a good settings

[cinder wyvern]

can we still use the applio google colab while this situation is going on

[rare goblet]

Ayo? @cinder wyvern level 5 !!!

[glacial pollen]

@clear citrus
https://ibb.co/GvdYmnv

Extra infer time = Extra
Sample length = Chunk
( for these two, refer to the link (( Image I've made )) ) - dw, it's safe. I've just lost perms to post images

N. Gate = noise gate Tweaking it is up to you to tweak it ( It's a threshold measured in Decibels. Anything quieter than your set value will be discarded )

Echo is anti-echo, sup1 and sup2 are noise suppressors.
These 3 elements do degrade audio most of the time and aren't perfect

[clear citrus]

i dont understand the link

[glacial pollen]

Well, just sit down and read

[cinder wyvern]

can someone answer me

[clear citrus]

alr

[glacial pollen]

Can't simplify it sadly
If you wanna understand what's what, you just gotta sit down and read a little. Gpt is helpful too for explaining certain stuff

[fleet swallow]

ok, figured that out but now it gives an error when inferring after 5 seconds

[fleet swallow]

ok apparently my audio file was too big for it to process

[brave garnet]

Voicechanger Download (Okada)

⠀
Download for Nvidia GPUs
Version 18a cuda

Download for AMD GPUs
Version 18a directml

Download for Intel GPUs
Version 18a directml

Download for Mac
Version 17b Mac
⠀

Made with love by Ilaria 💖

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 💖

[fiery raptor]

so can I ask for gpt sovits help here?

[brittle wing]

-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, 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

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

[fiery raptor]

is that for me?

[brittle wing]

Is there an native English speaker who can help me transcribing a rare song's lyrics

[shut sand]

Hello, does anyone know why the app keeps processing the audio infinitely? I wanted to use it with the vocals of the singer Ado and the process never ends.

[brittle wing]

Idk

[indigo garden]

kendrick lamar type ish

[brittle wing]

Nah

Kendrick Lamar is talented tho

[brave garnet]

RVC Colabs and Spaces

⠀

Local Forks 🖥️

⠀
Mainline RVC
Original project, suggested for advanced users,
by the RVC-Project team.

Applio
Simplified, suggested for all, by the Applio team.

RVC Studio
Simplified, suggested for all, by SayanoAI.

Mangio-RVC
Simplified, may not be supported anymore, by Mangio621.

AICoverGen
Simple yet great way to make covers, by SociallyIneptWeeb.

Replay
From the greators of weights.gg, excellent product for everyone.
⠀

Made with love by Ilaria 💖

[indigo garden]

what were u talkin about? ahah

[brittle wing]

It's a rare song I'm not a native English speaker I wanna make a lyric video

[indigo garden]

ah ok great

im actually a music artist myself, if u want a song to make a lyrics video you could try mine maybe

[brittle wing]

I just want the lyrics to the other song

[indigo garden]

which song

[brittle wing]

I must send you the file in private and all the lyrics I've already transcribed
Bluejay rising - ELEVATE

Or just tell me which index rate is good for inference and I'll leave you

[lost oxide]

guys how do i create a model

[inner cloak]

Hello, I am a French-speaking student seeking to convert my university books/lectures into audiobooks or "podcasts".
I've stumbled upon the coquiTTS on the internet, and both its multilanguage an low-specs capabilities are interesting me.

I've tested the default model (XTTS2, voice cloning with XTTS2, and French models) and it doesn't feel natural enough to be comfortably listened to for a long time.
I currently run coquiTTS on a laptop cpu (intel i5 13th gen), and I haven't got much experience in fine-tuning TTS models, so I was wondering if someone here could have some advices.

I came here since there seems to be a big community around TTS models, but I'm not quite sure what is RVC.
Maybe you know someone who fine-tuned a French model that I could use ? I don't really care which voice it has, I just want it to be a good narrator

[rare goblet]

Ayo? @inner cloak level 1 !!!

[odd shale]

You can start by reading the docs.

-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
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[odd shale]

If you would like to get a model done, you can make a request post in the #1159289738314919936 channel or read the docs above.

[inner cloak]

I don't want to get a model done, I just don't really understand what's going on here. Is RVC for TTS ?

[odd shale]

Nope.

RVC is a separate program, it's not TTS.

[inner cloak]

Oh, what do you guys use for TTS then ?

I guess you need a first generated voice to give it the proper tone

[odd shale]

Yep, Applio has TTS.

You can use a RVC model for TTS on Applio

You can know more reading the docs.

-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
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[inner cloak]

Thanks

[dense marsh]

-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, 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

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

[verbal dome]

What does this mean?

"(new_audioslicer) shawn@Shawns-MacBook-Pro-4 so-vits-svc % svc train -c "/Users/shawn/Downloads/AI Voice/so-vits-svc/configs/config.json"

[09:31:31] WARNING [09:31:31] warnings.py:109
/Users/shawn/Library/Python/3.9/lib/python/site-packages/urllib3/init.p
y:35: NotOpenSSLWarning: urllib3 v2 only supports OpenSSL 1.1.1+, currently
the 'ssl' module is compiled with 'LibreSSL 2.8.3'. See:
https://github.com/urllib3/urllib3/issues/3020
warnings.warn(

[09:31:37] WARNING [09:31:37] warnings.py:109
/Users/shawn/Library/Python/3.9/lib/python/site-packages/urllib3/init.p
y:35: NotOpenSSLWarning: urllib3 v2 only supports OpenSSL 1.1.1+, currently
the 'ssl' module is compiled with 'LibreSSL 2.8.3'. See:
https://github.com/urllib3/urllib3/issues/3020
warnings.warn(

[09:31:37] WARNING [09:31:37] warnings.py:109
/Users/shawn/Library/Python/3.9/lib/python/site-packages/urllib3/init.p
y:35: NotOpenSSLWarning: urllib3 v2 only supports OpenSSL 1.1.1+, currently
the 'ssl' module is compiled with 'LibreSSL 2.8.3'. See:
https://github.com/urllib3/urllib3/issues/3020
warnings.warn(

Traceback (most recent call last):
File "/Users/shawn/Library/Python/3.9/bin/svc", line 8, in <module>
sys.exit(cli())
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/click/core.py", line 1157, in call
return self.main(*args, **kwargs)
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/click/core.py", line 1078, in main
rv = self.invoke(ctx)
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/click/core.py", line 1688, in invoke
return _process_result(sub_ctx.command.invoke(sub_ctx))
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/click/core.py", line 1434, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/click/core.py", line 783, in invoke
return __callback(*args, **kwargs)
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/so_vits_svc_fork/main.py", line 128, in train
train(
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/so_vits_svc_fork/train.py", line 88, in train
datamodule = VCDataModule(hparams)
File "/Users/shawn/Library/Python/3.9/lib/python/site-packages/so_vits_svc_fork/train.py", line 44, in init
self.batch_size = hparams.train.batch_size
AttributeError: 'HParams' object has no attribute 'train'
(new_audioslicer) shawn@Shawns-MacBook-Pro-4 so-vits-svc %

[late charm]

Hey, can someone help me create an effect in Bandlab so that the voices don't sound so cracked and robotic?

[scenic dew]

frequent errors occurred. please check if the model of the framework being targeted is loaded problem

[rapid crane]

hello, does anyone know how to use MDX-NET kuielab_a_vocals on UVR 5? or is there other way to separate 2 vocals in one song?

[ocean quail]

How to turn a google drive link to a hugging face one?

[knotty moth]

simply download and reupload it to your huggingface account, read the guide:
https://rentry.org/fdg_guide_newer#uploading-the-files-to-huggingface

[alpine pivot]

how do you parse f0_file as None in an API call? i'm getting An error occurred: expected str, bytes or os.PathLike object, not NoneType, the recorded API call is showing as None but i guess i have to put something

[wide torrent]

The http start keeps crashing

[knotty moth]

-svc

[azure marsh]

https://cdn.discordapp.com/attachments/1159822429137412177/1194008385113301003/svc.gif?ex=65c13ef6&is=65aec9f6&hm=ce1f57bc316793f641c80f9b359ba9db7d1312af6918b1aed9c8468c74035bac&

[knotty moth]

it means you should stop using it being older than the recent RVC v2, as well as no more svc models to find among RVC models in: #1175430844685484042
read the guide for more: https://docs.aihub.wtf/

[orchid glade]

pretty new to this and maybe a dumb question Is RVC the only tool I can use to train models or is there other better programs

[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 inclued 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 💖

[ocean quail]

I need help the hugging face link is missing “?download true” in it I followed the steps

[wide torrent]

Hey pip for some reason isint downloaded and I'm using python 10.0.0 so idk

[rare goblet]

Ayo? @wide torrent level 1 !!!

[wide torrent]

It's saying that pip is not recognized

[wide torrent]

Bro I need help

Pip us annoying

Is

[brave garnet]

RVC Colabs and Spaces

⠀

Local Forks 🖥️

⠀
Mainline RVC
Original project, suggested for advanced users,
by the RVC-Project team.

Applio
Simplified, suggested for all, by the Applio team.

RVC Studio
Simplified, suggested for all, by SayanoAI.

Mangio-RVC
Simplified, may not be supported anymore, by Mangio621.

AICoverGen
Simple yet great way to make covers, by SociallyIneptWeeb.

Replay
From the greators of weights.gg, excellent product for everyone.
⠀

Made with love by Ilaria 💖

[strong flax]

hi, if I have an AMD card, which version is better to download so that the voice does not lag, please share (I know about the onnx file, but it still lags for me)

[jovial kiln]

Can anyone help I just downloaded and I’m trying to use it but whenever I talk there’s and echo that plays like the reverse of what I’m saying and the ai voice picks the echo up so it sounds really wierd does anyone know how to fix it

[odd shale]

You can try and read the docs.

-rt

[azure marsh]

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

[odd shale]

Plus, try and use the Deiteris' W-Okada fork.

[odd shale]

If you would like to use Deiteris' W-Okada fork (recommended over OG version) for AMD, just take a look at this guide and scroll to the "download" section.

https://rentry.co/ForkVoiceChangerGuide

[severe sand]

Am I understanding correctly that I can use these inbetween saves as pre-trained model to continue training from there?

[polar crater]

I already run Feature Extraction. What can i do

[knotty moth]

• make sure you have run preprocessing and feature without issues at all
• make sure the dataset zip has a folder with readable wav files
• try re-export the wav files using Audacity, or if it can't be read, try open in another editor (izotope RX, etc) with option to include "non-audio"/metadata disabled

[severe sand]

Also does anyone know why Tensorboard only shows a data point every 200 steps? can that be changed?

[rare goblet]

Ayo? @severe sand level 6 !!!

[polar crater]

I already done. https://docs.google.com/document/d/1XuxQYiqEhYrdYeCZRRLrmV_ciMKo0bV-jTCGHu_-5Cc/edit i do with this

[low shard]

run feature extraction again, at the end you would find a 'no-feature-todo' which means there's something wrong with ur dataset

are all ur files .wav? are there no special characters?

in case i'd suggest u to re export ur dataset as wav with audacity

[wild bobcat]

random question about rvc

i made a voice model of boyfriend from fnf and i used a chromatic scale on it

and some of the vowels sound like theyre being mixed together

why does it do this?

the first sound is supposed to be a ooo sound but it sounds like theres a "e" in the background for example

like its being mixed together

[oak sentinel]

I have trained my voice model using the RVC web ui and it now lives in ./assets/weights/myvoice.pth. How can load this model in a simple python app to perform text-to-speech? Does anyone have some functioning sample code?+

[late juniper]

Applio has TTS built in

[oak sentinel]

Is Applio a library or an alternative to the RVC web ui?

[late juniper]

It’s a better alternative to RVC web ui

Way better

[oak sentinel]

It looks deprecated. The repository has been archived. Are you sure this is sustainable?

[rare goblet]

Ayo? @oak sentinel level 1 !!!

[late juniper]

O shit I sent the wrong one

[rare goblet]

Ayo? @late juniper level 2 !!!

[late juniper]

https://github.com/IAHispano/Applio

It’s this

[oak sentinel]

thanks.

the trained pth file should be usable outside of these web uis by mere python, right?

[oak widget]

how do i make voice models of a friend or someone i know 💀 i'm new asf to AI
(any tutorial will be appreciated 🙏 )

[rare goblet]

Ayo? @oak widget level 2 !!!

[oak widget]

how many clips should the dataset contain

and how longshould each clip be

do you have any step by step videos or sm

[severe sand]

Is there a guide or something to the tensorboard graphs? so far I have only heard about the loss/g/total

[pastel oak]

#✨│ai-help message

[oak sentinel]

Does anybody have some sample code of how to use an RVC model directly in Python without any web uis?

[oak widget]

how can i make a modelll 😭

[severe sand]

Thanks for the reference.
In the screenshots they are comparing with 20 steps precision, my tensorboard only spits out datapoints every 200 steps.
I've searched wide and far but can't find anything about that, do you know if and how I can change this? or if its even relevant

[pastel oak]

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

[oak sentinel]

I found this video extremely helpful: https://youtu.be/hB7zFyP99CY?feature=shared

[pastel oak]

i dont think its thaaat deep, codename just likes his stuff very very very thorough compared to majority, youd be fine with 200 steps

i havent rlly looked into analysing graphcs like that to answer how to change that tho

[oak widget]

i know this sounds lazy and annoying but are there any instant websites in which i could just input the audio clips from someone and a song and it'll be complete?

[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! 💖

[pastel oak]

that

[knotty moth]

stop spreading misinformation through that old video

[oak widget]

wait so i could just input someones voice and a song and it will play it?

without a model?

[oak sentinel]

What is misinformation about it?

[knotty moth]

sovits-svc and it even sounds horrible compared to the recent RVC v2

-svc

[azure marsh]

https://cdn.discordapp.com/attachments/1159822429137412177/1194008385113301003/svc.gif?ex=65c13ef6&is=65aec9f6&hm=ce1f57bc316793f641c80f9b359ba9db7d1312af6918b1aed9c8468c74035bac&

[oak sentinel]

I dont get it. The video is not about so-vits-svc.

@knotty moth do you know if its easily possible to use an RVC model directly in Python without any web uis?

[severe sand]

Does RVC use 16k hz to train the model? or what exactly is this for?

[oak widget]

i uploaded the model and an audio but it gives me an error

[quartz heart]

where i can create covers for free ?

[knotty moth]

you can code your own using the functions for inference & training from the rvc code (i.e. infer-web.py)

[oak sentinel]

Yes, I am currently looking at it. Seems rather complicated though.

This seems promising: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion (not the Web UI, but an API). Any experience anybody?

[oak widget]

are there any instant websites in which i could just input the audio clips from someone and a song and it'll be complete

[oak sentinel]

I only know https://elevenlabs.io/

[rare goblet]

Ayo? @oak sentinel level 2 !!!

[oak widget]

elevenlabs doesnt have singing

i think

[oak sentinel]

they have an upload to change the voice of an audio clip. if you split the vocals from instrumentals prior, it might work.

[oak sentinel]

Why does Applio look for Discord?

[glad zealot]

its for rpc, basically just shows that you are using applio on discord

same thing as when you're playing a game, it will show what game you are playing on discord

[oak widget]

if anyone has any suggestions i would be grateful

[oak sentinel]

@glad zealot into which folder of Applio do I have to copy my voice model so that it becomes available in the Web UI?

[glad zealot]

i forgot what the applio file system looks like but i think it should be in the logs folder

the files should have the name similar to <your model name>-s<steps>-e<epochs>

i forgot

[severe sand]

Can you manually stop training when you notice that its starting to overtrain? obviously you can just exit out but how do you get your final checkpoint then

[oak widget]

do you know how to use Ilaria rvc 😭 you seem like someone who would know

[brittle wing]

I have an issue with applio the uploaded audio for inference is unplayable

And the converted result too

[brittle wing]

How much index value is ok?

[glad zealot]

0

mostly depends on your accent most of the time

since index is pretty much the accent of the model

[brittle wing]

0.7-0.75-0.8???

Which one

[glad zealot]

i havent used that in so long idk..

[brittle wing]

Also

[glad zealot]

like actually 0

thats what i use at least

[brittle wing]

Why zero

[glad zealot]

disables index

[brittle wing]

https://colab.research.google.com/drive/1X8YR4Ruv7zzY8YAMPTfC7hkxqT_d4Q5d
Hybrid pitch extraction or rmvpe

[glad zealot]

but when i do eventually use it i use 0.4

[brittle wing]

I want accent

[glad zealot]

0.4-0.7 seems to be good

still depends on your own accent

[oak widget]

do you know the fastest and easiest way to make a voice model? like very beginner friendly

[rare goblet]

Ayo? @oak widget level 3 !!!

[glad zealot]

its not magic so you still have to try to match the model's accent yourself, index is just there to help make it sound more believanle

[brittle wing]

Mhm yes

[glad zealot]

not sure but i do use mainline gui

[oak widget]

how long would that take to make a model?

[glad zealot]

depends on how long your dataset

but if you really wanted to you can train it for days

[brittle wing]

Depends on your dataset length also look at tensorboard to avoid overtraining

[oak widget]

okay i'm so lost. i have no idea what a dataset is in this context

and i have like 2 minutes of someones voice

is there any way i could use those clips to make them sing a song or smth

[brittle wing]

But of it's big it will take long to complete an epoch

[glad zealot]

thats just the audio file you have of the voice

[glad zealot]

it shouldnt take an hour to do that then

[brittle wing]

Samples of dry unreverbed Denoised vocal

[oak widget]

is the website straight forward? or do you have any videos ?

[brittle wing]

I mean lead vocals

No backing

[oak widget]

is there any video i can follow

[low shard]

https://docs.aihub.wtf/rvc/resources/datasets/

[low shard]

Nope, all videos are outdated

[oak widget]

so are there no websites that i could use to like input someones voice and input a video and then

[low shard]

Btw, all the google colabs of ilaria rvc are broken so dont use them, only the ilaria rvc zero works which is a zerogpu huggingface space but its only inference (use models)

[oak widget]

get them to use the words in that video?

for example a song or smth

im just trying to make simple AI covers of a friend or smth

are there no straight forward websites that are suited for beginners?

[low shard]

it doesn't really work like that

You need to train (make a model) first of that voice

What's your pc gpu?

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

[oak widget]

like

rtx 4060

[low shard]

laptop? how much vram? (seen in memory gpu in the method i told u above)

[oak widget]

PC, 32 gb

i think?

[low shard]

That's very good, you don't need to do it on cloud (online/site, like using a remote good pc) you can do it locally (on ur pc)

[low shard]

Are you really sure?

Please, check it

I told you the way to do that

[oak widget]

yeah alright

[low shard]

check it first then tell me cus its important

[brittle wing]

What's wrong w applio can't play audio

Nvn

It fixed

No

[oak widget]

it's 8GB i was wrong

[brittle wing]

How long is your dataset

How much dataset and epochs

[verbal dome]

whats the best alternative to so-vits-svc for making ai song using your own voice? The program is outdated apparently.

[rare goblet]

Ayo? @verbal dome level 1 !!!

[brittle wing]

How long is your dataset

[low shard]

That's still fine,

Okay so first of all:

1. Make the dataset
2. Download Mainline in the same page it will tell u how to use it

[verbal dome]

What?

[oak widget]

is a 7 minute dataset enough?

[brittle wing]

Ye

[low shard]

Yes

[brittle wing]

Well

[oak widget]

alright thanks a lot 🙏 can i dm you if there is a problem or would that be disturbing you?

[low shard]

If u need an issues u can just say it here, so you are always able to get help

[brittle wing]

Uh don't ask me I don't own a computer

Use Mel roformer denoise model by aufr

[rare goblet]

Ayo? @modern marlin level 5 !!!

[brittle wing]

On MVSEP

For free otherwise you have to pay if you use it on X-Minus/UVRONLINE idk if the dfiles were relewsed publicly

[knotty moth]

get better quality source

[brittle wing]

I use Mel roformer karaoke....it doesn't leave noise

It doesn't generate it's available on MVSEP

It's not it's for denoising

[white robin]

Hi I have a problem, as soon as I upload a voice in my voice changer it does not work and I have this message, I tried to install cuda and try on several voices but nothing works it always appears, a solution? 2024-09-22 19:14:42.5926927 [E:onnxruntime:Default, cuda_call.cc:116 onnxruntime::CudaCall] CUDA failure 35: CUDA driver version is insufficient for CUDA runtime version ; GPU=1730020260 ; hostname=MATHIS ; file=D:\a_work\1\s\onnxruntime\core\providers\cuda\cuda_execution_provider_info.cc ; line=125 ; expr=cudaGetDeviceCount(&num_devices);
*************** EP Error ***************
EP Error D:\a_work\1\s\onnxruntime\core\providers\cuda\cuda_execution_provider_info.cc:125 onnxruntime::CUDAExecutionProviderInfo::FromProviderOptions [ONNXRuntimeError] : 1 : FAIL : provider_options_utils.h:153 onnxruntime::ProviderOptionsParser::Parse Failed to parse provider option "device_id": CUDA failure 35: CUDA driver version is insufficient for CUDA runtime version ; GPU=1730020260 ; hostname=MATHIS ; file=D:\a_work\1\s\onnxruntime\core\providers\cuda\cuda_execution_provider_info.cc ; line=125 ; expr=cudaGetDeviceCount(&num_devices);
when using ['CUDAExecutionProvider']
Falling back to ['CUDAExecutionProvider', 'CPUExecutionProvider'] and retrying.

---

Sorry I can't send screenshots here.

[white robin]

i have a amd gpu

[knotty moth]

I'd recommend this voice changer: https://rentry.co/forkvoicechangerguide
make sure to download the dml installation

[white robin]

thanks

[rare goblet]

Ayo? @white robin level 1 !!!

[low shard]

seems fine

[orchid glade]

will lowering the batch size when training affect the quality at all ?

[rare goblet]

Ayo? @orchid glade level 5 !!!

[magic basin]

yo

my svc fork thingy crashes whenever i put in a config file

any fix?

[knotty moth]

read my notice here: #✨│ai-help message

[magic basin]

ty

so i should switch voice changers?

[ionic jolt]

No

Do I need to cut audio into 10 second clips or can I just put in something with constant talking

[proper shale]

No need to cut audio in 10 second clips.

Just make sure to remove silence

[alpine pivot]

bump

setting it to an actual f0 .txt works but i get a 403 error afterwards

[night rune]

-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, 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

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

[night rune]

What happened to easy gui again?

[candid lodge]

how to conver rvc models a onnx for usage amd gpu

[rare goblet]

Ayo? @candid lodge level 1 !!!

[low shard]

#📰│dev-updates message

[severe sand]

:/

[night rune]

It was literally the easiest notebook for training voice models

[low shard]

it wont ever be back

[night rune]

[low shard]

I'd personally suggest to use Kaggle Mainline as it gives 30 hours weekly of free gpu so no disconnection

but its harder to use

[night rune]

I don't understand how to use kaggle

[low shard]

there's guides like https://rentry.co/RVC-Mainline-Kaggle

Else if u care about it being easy, try Applio or RVCDISCONNECTED

[night rune]

Okay, I'll try one of those options, thanks.

[low shard]

yw

[iron cradle]

how do you make good ai covers

[bitter jetty]

A lot of data training and less reverb and echo

[rare goblet]

Ayo? @ionic jolt level 1 !!!

[ionic jolt]

Wait now I'm confused does batch size affect the quality for example I put the batch size to 4 instead of 16 will it be better. What's the best batch size if data set is more than 10 mins sorry for the question

[rocky summit]

Where can i download the ai voicechanger software?

[cinder grotto]

Question: Have they already uploaded the latest version of Applio to Colab, or is there a link to the latest version if possible?

[alpine pivot]

is there something i'm missing trying to use the applio "use via API" to generate voice clips? every time i try to use it i get
Loaded as API: http://127.0.0.1:6969/ ✔
server rejected WebSocket connection: HTTP 403

[rare goblet]

Ayo? @alpine pivot level 1 !!!

[alpine pivot]

not really sure what's going wrong

[pearl bramble]

how to fix

[rare goblet]

Ayo? @pearl bramble level 5 !!!

[pearl bramble]

so all the huggingface stuff is getting this

[cinder grotto]

people How can I fix this

[lean owl]

-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
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[rare goblet]

Ayo? @lean owl level 12 !!!

[lean owl]

-hf

[azure marsh]

<:huggingface:1179800228946268270> Hugginface Spaces

• UVR5 UI, by Eddy and Ilaria Huggingface Spaces
• Ilaria RVC Zero, by thestingerx Huggingface Spaces
• RVC⚡ZERO, by r3gm Huggingface Spaces
• Applio, by IA Hispano Huggingface Spaces
• 🆕 FaceFusion UI, by Nick088 Huggingface Spaces

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

[lean owl]

what happened to ilaria rvc on huggingface

is the site broken or something

[leaden crag]

For me it says 'Internal Error' so it's likely just a temporary issue that'll be fixed tomorrow 🤞

[lean owl]

ah

[cinder sundial]

I’m trying to make fnf chromatics and music but they suck does anyone think they could make me one?

[jovial kiln]

Where do I find those?

[rare goblet]

Ayo? @jovial kiln level 1 !!!

[frozen ledge]

I am having an issue with batch rendering with Mainline RVC where it will do a few samples and then stop

[magic badge]

-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, 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

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]

what is a good vc app that does not lag for downloadable voices?

[alpine pivot]

bump, anyone have ideas for this?

my program is about done now and i kinda just have the tts -> rvc voice integration left

but i can't figure out this dumb problem :p

i’m mainly just using applio because of familiarity and since i use it for general RVC stuff but i guess if there’s another option with an API if this is just impossible i’ll look into it

if i get one of the parameters wrong it will give an error based around the api call’s parameter list but

once the parameters are correct it’s just always a 403

[azure marsh]

Suggestions for @candid lodge

💻 (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

☁ (Online) W-Okada - Colab Version

• How to use Original W-Okada's Voice Changer Google Colab
• Modified W-Okada's Voice Changer Google Colab
• Original W-Okada's Voice Changer Kaggle
• Modified W-Okada's Voice Changer Kaggle

⚠️ Notes

• Colab free plan GPUs tipically works for about 4 hours each day
• Kaggle restricts GPU usage to 30 hours per week
• These options may not work on mobile devices due to the lack of a Voice Audio Cable (VAC)

❔ Russian FAQ for voice changer

• 🔗 ultimate-voice-models-FAQ/Voice-Changer

Credits: blackcat76it, elektroeban, redoverflow and tntmaxim

[low shard]

hf spaces seemed to have problems, looks all fixed now tho

[low shard]

https://docs.aihub.wtf

[low shard]

hf spaces seemed to have problems, looks all fixed now tho

[low shard]

hf spaces seemed to have issues, retry now, also i suggest u to use ilaria rvc zero which is faster

[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! 💖

[low shard]

whats ur pc gpu

[lean owl]

Ye it works now

[low shard]

hf spaces seemed to have problems, looks all fixed now, retry

[low shard]

[low shard]

outdated colab don’t use that.
Are you looking for inference (use model) or train (make model) ?

[pearl bramble]

do you know why all the hf spaces went down

[low shard]

it was just an huggingface space servers issue, can happen sometimes

[pearl bramble]

oh alr good that it is not gone forever

[low shard]

yup

[low shard]

oh wtf, you sure the tensorboard is refreshed and the smoothing is to max?

#1159290752195633273 , everywhere else it will be deleted

[low shard]

@surreal tulip alr just deleted it in the other channels which deleted also the message u replied to this, please dont ever send it in other channels again

[surreal tulip]

gonna repeat it again "i had no idea" so like duhh??? im not gonna do it again

[low shard]

alright, i just re-said it as it also deleted your reply to my message

[surreal tulip]

pretty sure the first time and the warning was enough, dont need a 3rd statement!

[low shard]

wtf never seen the graph like that, I seriusly have no idea Sorry

[brittle wing]

Why is there an empty field when I'm training the index?

Why is there an empty field when I'm training the index?

[low shard]

oh that’s more visible, looks like a model collapse

what did you set as the batch size?

[knotty moth]

tbh that makes you seem like an ad spammer

[low shard]

well you can just put a valuve over 1k, but its really not worth it as more epochs dont mean better

[brittle wing]

Nick help pls

[brittle wing]

.

[low shard]

So you click it and get nothing, not even in the cmd?

[brittle wing]

yes

no

[low shard]

Are you sure no feature extraction run fine without a “no-feature-todo”?

[brittle wing]

Now it just says 'training' and doesn't show anything further.

[low shard]

Be sure its super clean, ig 4 batch size could be okay but there is something wrong in the dataset

[brittle wing]

.

[low shard]

this when you click index? Try to wait if its training

thats the problem then, even if quality nd quantity are both important, quality is a bit more, be sure to clean it better

[brittle wing]

but it should show up immediately

And mine says 'training'.

I don't have the first epoch checkpoint in the 'weights' folder."

Nick088

@low shard

[low shard]

that depends on what u ut the save frequency

[low shard]

is it still saying training?

[brittle wing]

no

[low shard]

Then its fine

[brittle wing]

what next?

Because I don't have the first epoch checkpoint in the 'weights' folder.

[low shard]

If you didn’t do it yet, train the model and monitor the tensorboard

[brittle wing]

.

[low shard]

Because you need to train, and also the first epoch depends on your save frequency

[brittle wing]

So what should I do now if I don't have it?

[low shard]

You need to train it

[brittle wing]

press Train yes?\

yes?

[low shard]

what guide are you followijg btw?

[brittle wing]

https://rentry.co/rvc-mainline-colab

[low shard]

did you manually sync the graphs? https://rentry.co/rvc-mainline-colab#manually-syncing-your-graphs

[brittle wing]

So, I did the train index, but there's no first epoch checkpoint in the 'weights' folder.

[low shard]

did u do this too?

[brittle wing]

yessssssssssssssssssssss

helpppppppppppppppppppppppppppp

[low shard]

Then you should click that “train model”

[brittle wing]

TRAIN MODEL?????????????

[brittle wing]

..

Error

[brittle wing]

Why are you so slow in responding and helping?

[low shard]

I'm literally exiting out of school...I'm not an AI and its not like its been hours since the reply, its just mins

[low shard]

check what the output colab says

[brittle wing]

\

[low shard]

You sure you are on T4 GPU runtime?

[severe sand]

I dont really get how you pick out the best model when the loss is tracked in steps while the models are saved by epochs. Like when 6000 steps was the best but you only have saves in the +-500 steps range, wouldn't you want to get THE best if possible?

Wouldnt it be better if the Tensorboard datapoints equate to the steps at each epoch?

[late sage]

merhaba

[brittle wing]

What is a good free non laggy real time voice changer?

[severe sand]

So my graphs look like this but the last checkpoint is a little closer to the original than the 8800 one (lowest loss) what gives? I thought low loss basically means how close it got to the original. Maybe I'm also just imagining it since the difference is pretty small

[wild bobcat]

out of curiousity i decided to make a voice model using human speech pitched down an octave to see what happens

did the voice somehow like decide to default to one of the pretrains voices or something

?

im curious

[unreal spindle]

Hi I'm trying to clone my voice and I don't know which app or open source is good for this. I want it to be sound like real. I have my 5 hours clean audio file with impressions and reading books etc. How can I train my voice and take rvc .pth file?

[low shard]

What's your PC GPU?

[fluid flare]

chat help me

i made my voice model

and when i use it

my voice ai crashes

and when i manage to use it, it only like sounds ZZZZZZZZZZZZZZZZZZZZZZ

[low shard]

are you using voice.ai? its garbage

[fluid flare]

my voicemod doesnt work

[low shard]

you made an RVC Model right? What did you follow

[fluid flare]

what rvc model..

[low shard]

We don't use voicemod/voice.ai here, its paid and garbage

[fluid flare]

what yall use

[low shard]

We use RVC models, which are the best and free

[fluid flare]

do i install i

it

[low shard]

We use especially the program wokada for using RVC Models in realtime

[low shard]

What's your PC GPU?

[fluid flare]

idk

its a intelcore i5

and nvidia geforce rtx

[low shard]

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

[low shard]

That's the cpu

[low shard]

which

[fluid flare]

dont kno

[low shard]

.

the gpu matters much

[fluid flare]

i cant open it

the task manger

[low shard]

You are on windows right?

[fluid flare]

it says im not allowed to

and i dont have specified perms

[low shard]

What ? Are you maybe on a school pc or smt

[fluid flare]

no

my own pc

i think i got virused

😭😭😭

[low shard]

Wtf did you download?

[fluid flare]

idk

[low shard]

Cus you should be able to use task manager..

[fluid flare]

i mean i tried to download gta 5 superman mod

and uhh

didnt work well

i cant even open gta 5 now

[low shard]

Its way better you do a full scan for viruses 😭

[fluid flare]

i cant even scan

or open folders i install

i can only open folders if they are not abt changing smth on the pc

🥶

is it a virus

[low shard]

About changing smt on the pc?

[fluid flare]

or is it my pc buggin

[fluid flare]

yea

like CRU

for custom resolution

or a virus scanner

[low shard]

You mean like when you get the poup saying this program will modify the pc? usually happens for anti viruses

Is it a family pc maybe ?

[fluid flare]

no

[rare goblet]

Ayo? @fluid flare level 2 !!!

[low shard]

I never really heard of having problems with opening task manager unless it's a controlled pc (like by school) or got a virus

[fluid flare]

it js says '' you dont have the specified permission ''

[fluid flare]

virus?

how i remove it

i cant even open cmd prompt

[low shard]

I might be wrong, but Did your pc have any other weird behaviour?

[fluid flare]

yes

whenever i restart it my screen js stays black w a cursor for 10 mins

then it comes to normal

[low shard]

Is it an old pc? But still weird it does that

[fluid flare]

no

its new

i got it last yea

and when i try to screenshot something my whole taskbar just freezes

and after 10 mins it pops up to screenshot

[low shard]

This all happens only after the time you downloaded that superman mod ?

[fluid flare]

yes

[low shard]

I personally use Avira, Malware bytes and Bitdefender, you have any anti viruses that you can use ?

[fluid flare]

no

anything thats about scanning or smth i cant use

[vestal spindle]

(Repost from general, also sorry for interrupting your conversation)

I need...

I need help..

Not just any type of help...

I got a 4060ti (before I had a 1070) and I need to know what I should set my RVC settings to now

Mainly I am looking for sample length, fade length, and extra inference time settings

These are the settings I had on my 1070

[low shard]

Not even windows defender did anything?

[vestal spindle]

Disregard, fixed

[unreal spindle]

rtx 4050 6 gb

[low shard]

Not really the best as I heard its bad and the recommended vram is 8+, But if you really wanna do it locally (on ur pc) you can choose between:

• Applio: A fork of RVC with some extra features like Applio TTS, same quality tho
• Mainline: The original RVC

[unreal spindle]

Thanks for help and what if I want to use google colab? I can get more GPU power from there

[rare goblet]

Ayo? @unreal spindle level 1 !!!

[low shard]

All the cloud (remote good pc) ways are:

• 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)

which i personally suggest mainline kaggle as kaggle gives 30 hours weekly of better gpus compared to google colab 4 hour daily that could get you disconnected

[fluid flare]

no

[low shard]

What seems like to me is that you don't have admin perms on your own pc, there are alot of viruses it might be hard to find, hopefully its fixable without resetting the pc

[fluid flare]

i cant even factory reset it

[low shard]

[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, 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

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

[vestal spindle]

Hey

I'm back again

So basically

I am finding that the RVC when I have it turned on and recording in audacity to test

it seems to skip a couple times here and there

any idea why and how I can fix it? much appreciated thanks

[weary elm]

i'm using RVC1006Nvidia, where do i put .ckpt files?

[bitter jetty]

I've been trying to train my ai, so would it better if I included the thing I am trying to replicate singing to help them sing better?

[honest plinth]

does anyone have a video link how to do any of this?

[low shard]

theres no video, and do what?

[low shard]

you should not put parts of other voices in the dataset, there should be only 1 voice

[bitter jetty]

Ok thank you

[rare goblet]

Ayo? @bitter jetty level 1 !!!

[low shard]

yw

[knotty moth]

only clean dry voices to include, unless you want to make some instrument model, etc.

[bitter jetty]

Understood, thank you

[violet pecan]

yo gang how do i download the program

send proper link please its urgent i wanna prank my friendf

yo gang

i cant launch the program what do i do

[violet pecan]

nvm

found it

but kinda need help setting it up

[tame mica]

-rt

[azure marsh]

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

[hard jay]

-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 Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline Kaggle by Cauthess

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

[oak schooner]

Hello.
I’ve got a question.
Is it possible to «retrain» model from English to Japanese as an example? To get rid of accent and similar things.

[steel forge]

If you have the dataset, you can retrain with a different embedder. But no matter what, Japanese will sound a little weird on an English dataset

Turning off / adjusting the index might help

[oak schooner]

Hm… okay, thanks. True, but I want to try to reduce effect as much as possible.

Maybe I’ll just ask in another way
I have English dataset and is it possible to work with it? Maybe somehow TTS it to the Japanese or something..

[rare goblet]

Ayo? @oak schooner level 1 !!!

[solar token]

Yo

Minimum how many ram to use this thing?

Cuz it's quite lagging ngl

I got 3.20 Hz processor

Or 3.50+ idk

[stiff idol]

guys where do i find the voices so i can download them and use them

[lunar dock]

Hello, good morning! I have a lot of doubts about installing the voice changer, could someone help me? I'm used to just installing it in a casual way that already comes with an installer.exe.

[knotty moth]

that sounds like what some mc redstone circuits may achieve

[verbal dome]

Is this website legit? https://docs.aihub.wtf/rvc/resources/datasets/

[knotty moth]

yes, as pinned in this channel

[brittle wing]

I need help, i can't download nothing

[wary sierra]

Wdym

[brittle wing]

nvm

[rare goblet]

Ayo? @brittle wing level 1 !!!

[blazing igloo]

how do I create my own ai voice

[crude bolt]

get clean voice audio first - 15 minutes of ultra super mega clean voice sound - no noise, no krickets, no burping, snorting, giggling.

of 1 person

[wary sierra]

Is there any way to identify overtraining without the tensorboard graph? I've been watching the values in cmd for about two hours now, but I'm unsure of how to analyze them

[crude bolt]

no

[wary sierra]

so I'm practically gambling rn

[crude bolt]

correct

even knowing overtrained or not , is not necesarely worse or better , only way to indicate THAT , is just listening to it how it performs.

BAD - like quality , means nothing if the voicechangers SOUNDS like an angel , but say the wrong words.

example you say - bottle of water , and it says : battle of wetter

[wary sierra]

meaning the only thing I'm possibly losing is the amount of accuracy

[rare goblet]

Ayo? @wary sierra level 1 !!!

[crude bolt]

it is a mix of alot ingredients , where tensorboard gives an indication like will spending more time of training actually yield any progress in the result yes or no ?

some people even dont like diminishing returns , but that whole graph is so smoothed out , that you apply more smoothing and that upward trend is smoothed down again .... this is like just witch craft - how far you gonna smooth the average to see the turning point even - hence they did smoothing of 0.987

compare epoch 300 with 400. if its going worse , try epoch 500 ... if it even worse than 400 , than you know somewhere maybe 350 ... if sounds worse go even lower than 300. This doesnt say anything bout overtraining , but just if it sounds better or worse by ear and randomly trying to find the best training result.

even what you find sound best is also very subjectively.

[knotty moth]

more likely due to lack of dataset length and variations

[crude bolt]

imo variations > length.

if the 1 person can talk multiple languages you have so much more variations than only like training english 🙂

[verbal dome]

Can I use this website for Mac?

https://docs.aihub.wtf/essentials/whats-rvc/#what-are-the-requirements-for-rvc-locally/

[crude bolt]

press : apple key + F , type mac - i see 2 results

[verbal dome]

It says, " "Yes, on Macs of recent generations.
But you can only do inference & it's a little unstable." What does it mean by inference?

[crude bolt]

i have no personal experience with mac and rvc , i HAD a macbook pro 2011 2013 2018 - with 10 years of experience with mac - i will say I would be able to do it , and i think it is a pain in the butt to get it working or even performing well.

[wary sierra]

I'm an idiot, I just installed tensorboard and it works ... so sorry for wasting your time 😬

[crude bolt]

then again - i have no proper experience , lets say even if you get it working i dont know what software you need to use to route it through from rvc to discord , .... yes i had soundflower in the past and that was a piece of !@#*.

i would like end up giving up , try on parallels or bootcamp and then boot windows up 😂

things like these is really not 'the apple- way' of doing things - so i dont expect it to cooperate

[verbal dome]

Ah okay. I had a PC but it broke a few months ago, i brought it into a shop to swap out the keyboard and then it started having problems lol

[crude bolt]

since its put out this way - i dont expect it to perform live-rvc , but like transforming well audio file in , generate , audio file out.

[crude bolt]

for now our solution basically is king of the hill a nowadays windows 11 computer with an decent nvidia card installed.

[verbal dome]

Which would be better Mainline, Applio, Mangio, or AICover Gen? What I want to do is, make a voice model of my own voice and make it do AI song parodies.

[crude bolt]

nice goal. break it in babysteps. Your question is too big to have this 1 solved in 1 go.

can you sing?

[knotty moth]

!howtoask moment

[verbal dome]

No, i just want to send voice files of me parodying movie songs to my friends

I'll use Applio., it has the least cons on the website.

[crude bolt]

focus on that first. Try with autotuner - the voice is not the only problem while making parody songs.

You also maybe a bit expecting rvc to turn out well with your own created voice in the first place.

As expectation of your first voice created, your first attemps will sound bad. You will have to figure out why and how to make it better.

[knotty moth]

yea it has plenty features

[verbal dome]

Oh dang, I don't see a Mac Version. Maybe I missed it.

[rare goblet]

Ayo? @verbal dome level 2 !!!

[knotty moth]

the best bet is sing it by your own and convert it using the voice model

[crude bolt]

thats why i asked if he could sing 😛

you cannot expect applio to magically sing

[verbal dome]

I want it to be like the Spongebob parodies, but backwards. Like a the artist is singing a song but the AI just replaces the singers voice with mine.

[crude bolt]

you got the artist singing solo?

rvc will mess up if instruments are in the mix

[verbal dome]

I can extract vocals from a song.

[knotty moth]

don't bother on duets or group vocals if it's too hard to deal with it

[crude bolt]

before you create your voice , have you tried this with voice-models that you can download ?

[verbal dome]

I'd isolate the vocals, do AI magic, and then add the instruments back on. I have about 50 minutes of me speaking, so I just need to train my voice and get the AI program up and running and test out if it sounds good/works. If the AI is having trouble, I'll just record more audio of me but with different voice levels/pitches.

[crude bolt]

but ok i take it that you havent tried it with voice-models that you can download.

[verbal dome]

No, I haven't. Is there any benefits for me using Applio locally instead of the cloud? Or does it just do the same thing, but locally uses your computer specs to render.

[crude bolt]

i cant say that cuz personally no experience with rvc+mac , if i take 'mac' out the equation ; you have more grip of what going on.

[verbal dome]

This is the main website for applio, right? https://applio.org/learn/54

[crude bolt]

cant confirm.

(dont know)

[low shard]

Yes

The applio situation is all fine now

The fake applio was named applios.org

[knotty moth]

both local & cloud applio should be safe

[verbal dome]

The guide from their website for Mac gives a link to download Python but it's an older version of Python, is that fine?

[verbal dome]

Is there any cons to using their cloud version?

[brittle wing]

Will i need to install virtual audio cable in order to use the voices in discord or in audacity? basically elsewhere on my PC? or does this voice changer have some kind of built in system?

[knotty moth]

in discord yes, but non-realtime RVC is better for recording

[brittle wing]

Not too sure what that is, i'm on the github page and cannot figure out how i'm meant to find the downloads, it used to be neatly inside a box, now it's all over the place

[knotty moth]

better check the guide: https://docs.aihub.wtf/

[brittle wing]

I found this https://huggingface.co/wok000/vcclient000/tree/main

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

Someone said ONLY DOWNLOAD CUDA

[knotty moth]

for Nvidia gpus: https://huggingface.co/wok000/vcclient000/resolve/main/MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a.zip?download=true
for others: https://huggingface.co/wok000/vcclient000/resolve/main/MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.18a.zip?download=true

[brittle wing]

I have trust issues but i'll trust you

[knotty moth]

it is straight from that link

[brittle wing]

Ok TY, i think i should HOPEFULLY be able to figure out the rest, just trying to find the download was the hardest part for me

@knotty moth Reason i'm worried is because that's the ONLY ONE without the saftey tick no virus thing

Most of them have the shield apart from the one linked me which is why i'm suspicious

Also how do you know which one is for what GPU? they all look the same aside from the numbers

[rare goblet]

Ayo? @brittle wing level 2 !!!

[knotty moth]

the number is the version, but directml is for non Nvidia gpus

[brittle wing]

Ahh i see

[verbal dome]

Okay I ran the command "run-applio.sh" now what do I do?