✨│ai-help

[elder plume]

so add it to where this located?

[simple ore]

"C:\Program Files\AMD\ROCm\6.1\bin" ---> environment variables

[elder plume]

yes i did that, now its to browse

[simple ore]

err...5.7

if you did add manually, then stop browsing and click ok

[elder plume]

i cant seem to find it now

[simple ore]

if you have this already

change to 5.7

[elder plume]

okay so i found the 5.7

[simple ore]

for the love of god, click cance

[elder plume]

just press okay?

[simple ore]

click cancel

[elder plume]

okay

[simple ore]

and manually edit the value

[elder plume]

so what do i edit it into?

[simple ore]

i dont know how to make it larger

[simple ore]

yes

click ok, ok

[elder plume]

so that should be it for that part

[simple ore]

yes,

did you run patch zluda 5.7 bat?

[elder plume]

this right

the 57 one

[simple ore]

yes

[elder plume]

okay, lemme run it

[simple ore]

after that run-applio-amd

[simple ore]

go to train tab

[elder plume]

yes it successfully opened it seems

[simple ore]

at the bottom section expand the advanced settings

and click GPU settings

[elder plume]

okay, done

[simple ore]

does it show 7800 xt there?

or does it show radeon graphics (tm)

[simple ore]

okay, thank god it is done

[elder plume]

i thank you very much for this help

[rare goblet]

Ayo? @elder plume level 5 !!!

[elder plume]

and im sincerely sorry to cause such a headache

as im not the brightest when it comes to computers, especially software

was more interested in building pcs using physical hardware

@simple ore , thank you very much and im very sorry to cause so much trouble 😅 , you have a good day

[simple ore]

after you try using it you'll have a rare opportunity to spend 15-20min of your time on some quality self-education

[elder plume]

alright, ill do so

also, just a little embarrassing note, i am 19 years old and well it seems i got lots to learn still

but anyways thank you and i wont take up your time anymore

[sterile lichen]

why does it take 7-8 min to finish one epoch?

[simple ore]

depends on the size of the dataset.. maybe you running on CPU? 🙂

[sterile lichen]

nope, gpu on 100% so it can't be

[simple ore]

8 minutes/epoch is ~90 min set running on 3060

[sterile lichen]

largest file is 3.3Kb

[sterile lichen]

i got 1050ti, is that why?

[pastel oak]

training not recommended on that gpu

[sterile lichen]

yeah i figured, i'm gonna stop it at 50 epoch and see if it's any good if not i might buy Elvenlabs creator plan and do it there

[low shard]

11labs is TTS (Text To Speech), while RVC is STS (Speech To Speech)

It's kind of different lol

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

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

[sterile lichen]

they do have sts, not sure how good it is tho
thanks for the list i'm gonna check them out

[rare goblet]

Ayo? @sterile lichen level 2 !!!

[low shard]

Ah didn't know

Usually most sites (like weights.gg, kits.ai, jammables/voicify, etc) just use RVC, idk about 11labs tho as they developed their own TTS

[golden sphinx]

how do i train my model with gpu

[low shard]

what's ur pc gpu?

[golden sphinx]

none

i use the colab version

speaking of which, im out of time

how do i get more time

[low shard]

it's impossible your pc doesn't even have a gpu

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

[golden sphinx]

i do have a gpu, but it's an intel

so the app itself doesnt work on it for some reason

[low shard]

so like:
Intel UHD (numbers) or Graphics ?

[golden sphinx]

how do i make the app work with an intel gpu

intel uhd graphics

just graphics

[low shard]

yep that's integrated graphics, you can't

your pc is too weak for that locally

[golden sphinx]

😭
how do i get more time on the online one

for colab

without paying

[low shard]

When you finish your Free Daily GPU on Google Colab, you can:

• Use an alt google account
• Use kaggle which gives more gpu time but its harder
• Wait until tomorrow
• Pay for colab pro

Kaggle is the best option tbh

[low shard]

Kaggle, the only good way, gives 30 hours weekly but needs a phone number

Or else use an alt google acc, but u still can run out of time on that too as google colab free gpu time is not granted

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

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

[golden sphinx]

-rt

[azure marsh]

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

☁️ (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)

[fleet heron]

I SEEM to have installed and set up everthing correctly, I'm just having trouble with audio input/output, and thus can't actually know if I set things up correctly

[ionic canopy]

First, sorry if I am interrupting the conversation. But I have a RVC model of Gale Dekarios from Baldur's Gate 3, and I have audio that comes from a person who isn't british, but has a clear american accent. Is there a way to have make the voice conversion have the accent?

Since the voice itself was trained on actual, unbroken dialogue of Gale

[low shard]

so have the accent of the audio that will be inferenced?

because it should have the accent of the voice model (if the model has an added index file, which is suggested for more quality so most models have it)

If you want it to 'ignore' the accent of the model and go more with the one of that american person, you can lower or put to 0 the search feature ratio or index ratio feature (the name depends on what ur using)

[low shard]

Wrong help channel, use #🔍│help-w-okada or #1192011222023950368
I'm not a wokada helper but i can tell you to not follow yt tuts

[fleet heron]

its rvc tho

[rare goblet]

Ayo? @fleet heron level 1 !!!

[simple ore]

index search = 0 -> use pronunciation(accent) from the inferred audio, use index search 1.0 to use accent from the voice model, or something in between for a mix

[low shard]

Are you talking about the go-realtime.bat?

like realtime voice changer for calls

[brazen merlin]

Can someone teach me how to use this on Mac?? Also is it free? I need more one on one help to set up. Please help.

[low shard]

What are you talking about?

Be specific on what u tryna do

[brazen merlin]

Trying to make my own ai voice and then try to get the voice to say something with words that I type. I’m on Mac though.

[low shard]

So, train a model and use text to speech

RVC is made for Speech To Speech not really TTS tbh

If you are looking specificially for TTS

There are different Text To Speech (TTS) AIs:

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

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

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

With RVC Models:

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

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

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

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

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

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

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

I'd suggest using another program than RVC

[brazen merlin]

Hmm ok.

Which program do you recommend for me if I’m on macOS.

Is it free?

I wanna be able to make multiple voices.

[low shard]

Mac isn't really the best for AI tbh,
The best AI TTS would be gpt so vits but i don't think it has mac support, u could try tho

[brazen merlin]

Idk if you know who sbtk is but I wanna make winx ai voices like that.

[low shard]

https://github.com/RVC-Boss/GPT-SoVITS?tab=readme-ov-file#macos

It works on mac, but uses CPU, could risk being super slow as

[low shard]

No idk him

The best tts specific program is gpt so vits

[brazen merlin]

Ok.

[low shard]

If you want to use RVC models which ofc are more as its more popular, u can try applio, a fork so modified version, https://docs.applio.org/applio/getting-started/installation , but ofc it's not as good for TTS as RVC is not really made for that

[brazen merlin]

Could I send a vid to you as an example of what I want the voices to do.

[low shard]

I got what you are trying to do dw

It's called Text To Speech

[brazen merlin]

Oh ok. Can it make swear words.

Which would give me the best results for what I want to do.

[low shard]

Uh yes both programs don't really have any censors about that

[brazen merlin]

Ok.

I have a Mac Pro so hopefully my rigs are good.

[low shard]

The best tts in quality if you want emotion would be gpt so vits, but it's harder and uses CPU so might be slow

[brazen merlin]

Ok.

Where do you recommend getting the best voice models to train

[low shard]

So, you mean the dataset to train a model? Or getting the already made model by someone?

[brazen merlin]

Either. If I want specific dubs of a voice I might have to idk vocal isolate the voice and make it a mp3??

[low shard]

About datasets, that are what you need to train models, you need a good quality one (NOT mp3, it's better to use wav or FLAC as they are lossless) you can check https://docs.ai-hub.wtf/rvc/resources/datasets/

About already made models, You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models :
  • Send " @gusty kestrel search (name of the model)", without the ()
  • Do /find with @earnest musk
• https://weights.gg/ (login required)
• 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/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

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

[brazen merlin]

It says x code installed but I can’t brew install ffmpeg

In terminal

it says command line tools already installed use software update in systems settings to install updates?

[rare goblet]

Ayo? @brazen merlin level 4 !!!

[brazen merlin]

Wym level 4

[brazen merlin]

Why is it slow on mac? Also the steps aren’t working

[low shard]

because it uses CPU, which means it's slow

Using the Mac M chipset resulted in lower quality results so they temporarely are using CPu

Unfortunately I don't think i can really help you for that as I don't do things locally nor have a mac

[brazen merlin]

☹️ ok

[low shard]

@proper shale did u ever use gpt so vits on mac?

[brazen merlin]

The 3rd step I can’t seem to install conda create gpt sovits python =3.9

I run the brew install it says zsh: command not found: brew
but it doesn’t work this is the 2nd step I’m stuck on.

I did manage to install x code command line tools though.

[simple ore]

you need to isntall homebrew

[modern surge]

-colab

[azure marsh]

📒 Google Colab Notebooks

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

ℹ️ Note

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

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

[proper shale]

nope

[summer scroll]

Link donwload amd?

[placid wren]

I'm having an issue with finetuning, and I really could use some help! I do the entire process, but after I compress and move the files, the "wavs" folder is completely empty. I never get any errors or anything up to that point, but I simply can't get any audio files to show up in that folder. I have tried different voices and files, yet this always happens.

[rapid crow]

my "res" is very high, is there anyway to fix that

but only with custom models

[uncut mason]

Im not experienced with AI models at all but i was wondering if I could make a simple translator app with a 30k sentence dataset from my native language which there is no support for with translations to dutch. Ik it won't be super accurate but im curious about what result it would reproduce. r there any tips on how i could improve accuracy? currently this is the script i use and i let it run on google colab

from datasets import load_dataset
from google.colab import drive
from transformers import MarianMTModel, MarianTokenizer, Seq2SeqTrainer, Seq2SeqTrainingArguments

# Mount Google Drive to save checkpoints (optional)
drive.mount('/content/drive')

# Load your dataset
dataset = load_dataset('csv', data_files='e-translation_data_clean.csv')
train_test_split = dataset['train'].train_test_split(test_size=0.1)
train_dataset = train_test_split['train']
eval_dataset = train_test_split['test']

# Load pre-trained model and tokenizer
model_name = "Helsinki-NLP/opus-mt-nl-en"
model = MarianMTModel.from_pretrained(model_name)
tokenizer = MarianTokenizer.from_pretrained(model_name)

# Preprocess the data
def preprocess_data(examples):
    inputs = tokenizer(examples['source'], max_length=128, truncation=True, padding='max_length')
    targets = tokenizer(examples['target'], max_length=128, truncation=True, padding='max_length')
    inputs['labels'] = targets['input_ids']
    return inputs

train_dataset = train_dataset.map(preprocess_data, batched=True)
eval_dataset = eval_dataset.map(preprocess_data, batched=True)

train_dataset.set_format(type='torch', columns=['input_ids', 'attention_mask', 'labels'])
eval_dataset.set_format(type='torch', columns=['input_ids', 'attention_m```

ask', 'labels'])

# Define the training arguments
training_args = Seq2SeqTrainingArguments(
    output_dir="/content/drive/MyDrive/colab_checkpoints",  # Save to Google Drive or use local path
    save_steps=3600,  # Save checkpoints every hour (adjust as needed)
    save_total_limit=3,  # Keep only the last 3 checkpoints
    evaluation_strategy="epoch",  # Evaluate after each epoch
    learning_rate=1e-5,  # Lower learning rate for smaller datasets
    per_device_train_batch_size=16,
    per_device_eval_batch_size=16,
    num_train_epochs=5,  # Extra epochs to learn from smaller data
    predict_with_generate=True,
    logging_dir="./logs",  # Directory to save logs
    logging_steps=500,  # Log every 500 steps
)

# Define the trainer
trainer = Seq2SeqTrainer(
    model=model,
    args=training_args,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    tokenizer=tokenizer
)

# Check if a checkpoint exists and resume from it
trainer.train(resume_from_checkpoint=True)
# resume_from_checkpoint=True

# Save the fine-tuned model and tokenizer to Google Drive or local directory
model.save_pretrained("/content/drive/MyDrive/colab_checkpoints/fine_tuned_translation_model")
tokenizer.save_pretrained("/content/drive/MyDrive/colab_checkpoints/fine_tuned_translation_model")```

R there also recommendations for an alternative for google colab, i dont mind paying

[gloomy zephyr]

any way to stop them from doing the really high voice cracking thing during high notes? I think I've inputted enough high notes in the training but idk

[tame mica]

dataset issue i fear

you can try autotuning the input audio first to fit the range and note correct it afgter but idk

[brittle wing]

Uh tht happens w me too

[foggy fiber]

Does anyone know how i can Create Duet Song AI Covers?

i want to Use rewrite the stars from the greatest showman with different voices, i dont want to use Voices from Male Part Only (or Female Part Only) Karaoke Videos like from "Clark In Studio" or "BbKaye" ........ i want to Use the voices from the Original Songs, but i dont know how to Seperate Zac Efron's Voice and Zendaya's Voice, especially in the Harmony Parts

[latent kettle]

Use UVR to extract vocals

But for it. You require a decent pc if you don't have it you can use cloud services

[sudden lynx]

bro I can't extract RVC-GUI

that just stuck on 13 processed

[low shard]

RVC GUI is Outdated

Do NOT follow yt tuts

What's ur PC gpu? And what are u looking for

[sudden lynx]

Intel I7-10700KF and I want to switch lyrics to a song

[low shard]

That's a CPU not GPU

[spiral galleon]

help

[low shard]

To change the lyrics of a song you'd need RVC, sing the modified lyrics urself, get the instrumentals of the song, then convert your lyrics with the RVC model

[low shard]

!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.

[spiral galleon]

i am stuck on convert

[low shard]

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

[low shard]

Elaborate more, it's not an only program for everything

Say what are u doing and using, and be sure to Not follow yt tuts

[spiral galleon]

what do you mean?

?

[low shard]

What are you using?

What are you doing?

[spiral galleon]

RVC GUI I am Stuck On Coverting

[sudden lynx]

I did it

[spiral galleon]

it doesn't convert

[rare goblet]

Ayo? @sudden lynx level 1 !!!

[sudden lynx]

shut up

[low shard]

Rvc GUI is outdated

[sudden lynx]

so I did it I wrote my gpu

[low shard]

Don't follow YouTube tuts

[spiral galleon]

ok

[sudden lynx]

Nah so what should I do now?

[low shard]

What are you looking to do? And what's your PC GPU?

[low shard]

Intel i7 is a CPU, not a GPU

[sudden lynx]

oh

sorry

[spiral galleon]

i am i 5

[rare goblet]

Ayo? @spiral galleon level 1 !!!

[low shard]

It's fine

[spiral galleon]

i5

[low shard]

What

[sudden lynx]

rtx 3080

[low shard]

That's the CPU not gpu

[low shard]

Ay that's good!

As you got a good PC, you can use RVC locally, you can choose between:

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

[spiral galleon]

Intel(R) Iris(R) Xe Graphic

[low shard]

Ah that's very bad,
You can't do it locally (on ur PC), you have to use cloud (remote good pc)

So what do you want to do?

[spiral galleon]

convert

wait my pc is weak?

when i bought it it said it's an gaming pc

i got scammed?

[low shard]

Yep very weak

Unless
Do you have multiple GPUs, like GPU 0, GPU 1?

Bc the CPU is not that bad, but it seems that your GPU is pretty old

[low shard]

Ai covers right?

[spiral galleon]

yes

[low shard]

Could you send a screenshot of your task manager GPU tab? Maybe you got multiple GPUs

[spiral galleon]

bruh i got 1k+ dollars scammed

[low shard]

No way 💀

[sudden lynx]

I don't understand tbh I want to make ai voice like switch lyrics to a song remove the vocal and switching it with my words

[spiral galleon]

but i want to convert ai thing

bruh this pc is more cheaper then mine https://www.amazon.co.jp/Acer-Predator-第5世代AeroBlade-3Dファンテクノロジー-PHN16-72-N76Y46/dp/B0D1BDS6LD?source=ps-sl-shoppingads-lpcontext&ref_=fplfs&ref_=fplfs&psc=1&smid=AN1VRQENFRJN5

[low shard]

Yes and I explained you how to do it, first of all you have to download one of the programs I told you,
then sing the modified lyrics urself,
get the instrumentals of the song,
then convert your lyrics with the RVC model

You can't directly just type the words to do that

[spiral galleon]

nick you gonna help me?

[sudden lynx]

bro I can't sing it

I want to match the lyrics to the song

I can't do it?

[low shard]

I can't really read the Japanese specs, but I just want to tell you laptop GPUs are always weaker than a Desktop pc

[low shard]

Welp, there's no other way to do it with RVC

There's Text to Speech in Applio, TTS, but it's not really good for emotions or singing

[low shard]

Could you please send me a screenshot of your task manager?

[simple ore]

it is in english

laptop 4060 8gb vram

[low shard]

on my phone it opens the japanese page

but on my laptop the english one

wtf?

[low shard]

eh still a laptop gpu but ig

[shadow sail]

I'm sorry for the (most likely) very dumb question, but I'm encountering an error while using CoverGen_NO_UI_v2 and I tried several things to try and fix it, but it's not working for me since yesterday. Everything says it loaded fine, but when generating the cover it says "ModuleNotFoundError: No module named 'src'". Can you please tell me what I'm doing wrong?

[low shard]

send the link of what ur using

[shadow sail]

https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?authuser=2#scrollTo=xwmeRJL8PMaP

[shadow sail]

If you meant the model, I was using this one. Sorry, new to discord and to AI. ^^' https://huggingface.co/coivmn/hanbinnew/resolve/main/hanbin-coivmn.zip?download=true

[rare goblet]

Ayo? @shadow sail level 1 !!!

[low shard]

mm, lemme test it out

[simple ore]

doesn't colab require you re-run some cells/clone code after you DC?

[shadow sail]

Yes, I tried disconnecting, switching accounts, reloading everything there, and doing it again, but it's not working anymore for me

[low shard]

you mean after you get disconnect and run the cells again? yea ofc

But i just seen that there has been a PR approved yesterday that changed some things related to that 'src' so ima test it out

mm it took a while to extract the model

[low shard]

@viscid moss yea that PR seems to cause issues

I ran!git checkout 71b8ab6508e27c57b7336c5c605b1004d1c0a76f before the generate cover cell, to go to the latest commit before that pr

And it works

You just need to revert the latest PR

[low shard]

@shadow sail you can use this temp fix meanwhile

[shadow sail]

Thank you very much for all your help, I'll give it a try

[low shard]

You're welcome

[viscid moss]

Oh, Ryouko PR. I'll revert the changes when I get home

[low shard]

Alr

[brittle wing]

Someone please explain carefully what are the steps in fine-tuning an RVC model

[low shard]

you mean train your own RVC model?

[brittle wing]

No what I mean is I have already trained a model but I wanna start from the same checkpoint with a better dataset.How do I

[low shard]

You can just use the G & D files of the trained model as pretrains for finetuning your model, and then train like normal

I'm not really sure if u would wanna retrain or finetune if the previous dataset was shit tho

[distant turtle]

-colab

[azure marsh]

📒 Google Colab Notebooks

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

ℹ️ Note

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

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

[brittle wing]

Hm like resuming

But how

[low shard]

Well, u can't resume with a different dataset

Was the previous dataset bad?

[azure marsh]

Fine-tuning: Further improving an AI model, training it with a another dataset.

-# Source: AI HUB Docs

[simple ore]

1. you can continue training the same model with extra epochs

[brittle wing]

No ofc

[simple ore]

2. you can use D/G weights from the model as pretrain for another

[brittle wing]

Huh 🤔

[low shard]

Yea but I don't think he can do that with a different dataset

iirc all the settings and dataset have to be the same

[brittle wing]

Howww

[simple ore]

for 1) RVC has a bug where it incorrectly restarts and overwrites last weights/model

[brittle wing]

???

[azure marsh]

Overtraining: un modo efficace per rilevare il sovrallenamento è controllare se il TensorBoard Graph inizia a salire e non torna mai più giù, portando a un output robotico e ovattato con scarsa articolazione.

[simple ore]

wdym how, just instead of default f0d48k.pth put your d file

on the UI

[brittle wing]

Why I can't find the guides?

[low shard]

Or am I wrong? (talking about actual resuming not finetuning)

[brittle wing]

I said fine-tuning

[brittle wing]

Where

[low shard]

Which guides are you following?
that looks like the error page from https://docs.applio.org

[simple ore]

depends on what you're using for training

[brittle wing]

Rvc Disconnected

[simple ore]

rvc disconnected does not seem to have the path for pretrains

but you can drop the D/G files in the model folder

[brittle wing]

It has a drop-down menu

[simple ore]

pick custom

[low shard]

the issue is that it needs an huggingface links not directly the file tho, unless he uploads it to HF

[foggy fiber]

I meant I wanna extract Zac Efron's Vocals and Zendaya's vocals but also separate them apart too

[main mesa]

hi

can i have some help?

[remote steeple]

no

[low shard]

elaborate ur issue

[viscid moss]

Changes reverted, now it's working again

[low shard]

[odd shale]

[muted anchor]

where and how can you publish your finished voice model?

[low shard]

read https://docs.ai-hub.wtf/extra/model-maker-role/

[muted anchor]

Ok thanks

[rotund osprey]

Where do I put these settings?

RVC v2
RMVPE
500 Epoch
13000 Steps

[low shard]

those are training info,
I'm guessing you're using wokada and no u can't change those

the epochs and steps don't mean more quality, don't worry about those

the rvc v2 type is already selected

[rotund osprey]

thx

[low shard]

yw

[muted anchor]

why i can't upload the file?

[low shard]

try to refresh and re-upload

[muted anchor]

Ok

[cyan rampart]

How do i download this? rvc

[low shard]

what's ur pc gpu and what are u lookin to do

[cyan rampart]

to make ai songs

like ai covers

[low shard]

alr, whats ur pc gpu

[cyan rampart]

1660 super

nvdia

[low shard]

Your GPU is good enough to do inference (use models) locally (on ur pc), you won't be able to train (make models) but use them

You can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides):
  • Applio: A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
  • Mainline: The original RVC
• Cloud (remote good pc, easier and faster than ur PC but it's limited):
  • Ilaria RVC Zero: fastest and simplest that you can get for free
  • Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
  • Applio Colab: max 4 hours, not granted, of GPU

[cyan rampart]

i dont want to make models cause my pc is mid

but like i want to do covers ect

[cyan rampart]

this is an off topic question but do you know how this was made https://www.youtube.com/watch?v=khKdXTmbhDw

like 1950s type song idk

[low shard]

that can't be made with RVC

that's probably made with Suno or Udio, idk tbh

But with RVC u can make AI Covers

[low shard]

for ai covers u are all right with it

[cyan rampart]

ok

[low shard]

ai covers meaning changing the voice of a song, not the style like the 90s

[cyan rampart]

i dont understand what you mean withapplio and mainline like they are red

blue

https://github.com/Tiger14n/RVC-GUI/releases/ is this what i should be downloading

[low shard]

Absolutely not

super outdated

never follow yt tuts

[cyan rampart]

oh shoot

i should delete

[low shard]

mainline = original rvc
applio = modified version with a different interface that can also be kinda faster

U can choose one of them, applio might be easier

[low shard]

yep

[cyan rampart]

ok

is appolio better gui and easier than mainline?

[low shard]

applio*
And yea could be easier as you directly have to upload the files, while mainline you have to put the file path for basically everything

[cyan rampart]

ok ok

ill download it now

im trying to find the applio downloading

[rare goblet]

Ayo? @cyan rampart level 2 !!!

[low shard]

Compiled -> Linux or Windows (depends on your Operative System, i'm guessing you're on Windows) -> Click the download Icon next to ApplioV3.2.6.zip

[cyan rampart]

do i save it in my c drive yes

[low shard]

yea

[cyan rampart]

okok 4 gb ill wait and do it

[low shard]

alr

I may fall asleep as it's 1am

[cyan rampart]

ok

[turbid gyro]

how do i do ai text to speech w this

[turbid gyro]

hi

[cyan rampart]

Anyone know how to split the audio and instrumental

[ebon estuary]

Can anyone help me out? my epoch is stuck at 1: 0%, its either very slow or stuck im not sure.

https://i.gyazo.com/06b09a1b19ea72c4f44e3e132fbdceb9.png

[azure marsh]

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

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

[ebon estuary]

1070 gtx btw

[wicked shore]

guys i am very confused there are more download links then the video should i just get the newest one.

[hollow dome]

i feel stupid as hell but i feel like none of the colabs do inferencing and all of the huggingface spaces dont work
i dont want to train a model i want to actually use it how do i

[latent kettle]

You can use ilaria RVC it is very easy and good

[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! 💖

[latent kettle]

Use this link @hollow dome

[latent kettle]

For what?

[hollow dome]

I keep getting errors :(

[latent kettle]

What kind of errors

[hollow dome]

doesnt say

i figured it out nvm :) thankyu

[latent kettle]

Yw : )

[low shard]

btw, its better to not directly suggest cloud ways, as there are people who can do it locally with their good pc

it’s better to always ask the pc gpu

[hollow dome]

is there a colab or something else i could use so i dont have to wait?

[low shard]

for what?

elaborate more

[hollow dome]

i have an account though i am logged in

[low shard]

yea its your zerogpu quota

it’s not unlimited

[hollow dome]

mhm mhm

[low shard]

you can either wait or use applio colab

https://docs.ai-hub.wtf/rvc/cloud/applio-colab/

[hollow dome]

thank you very much

[low shard]

You’re welcome

[low shard]

btw it says that even when u have an account

don’t ask me why, its a zerogpu moment

[hollow dome]

odd lol

[low shard]

btw the applio colab should give u around 4 gpu hours daily for free (not granted), i think that should be enough lol

[hollow dome]

it def will be! thanks a bunch

[low shard]

your welcome

[elder plume]

@low shard , may i ask how to post a model to this server

as i recently made a model and i want to get it listed

[low shard]

read the link i sent in #1159514067187277865

[elder plume]

okay

[river tartan]

sorry first time here, got this problem when running RVC-GUI.bat
Running with the system Python .
Traceback (most recent call last) :
File "C:\Users-----\Desktop\MediaLibrary\3d Models & Projects\Projects\Ai sound\RVC—GUI—main\rvcgui.py", line U,
import soundfile as sf
ModuleNotFoundError: No module named ' soundfile'
Press any key to continue . . .

[pastel oak]

use a program more up to date

delete that and get this for example
https://docs.ai-hub.wtf/rvc/local/applio/

[river tartan]

ok thank you

[mild quiver]

which tts is best for human like responses?

[low shard]

There are different Text To Speech (TTS) AIs:

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

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

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

With RVC Models:

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

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

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

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

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

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

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

[mild quiver]

ty! guess i'll give elevenlabs a try

[low shard]

alr, yw

[brittle wing]

#outdated-model-maker-role

What happened to the channel where I submitted

[crude flame]

new thing

[brittle wing]

I- have to resubmit

Is it that one?

[crude flame]

yeah

i dont think the bot works atm though

[brittle wing]

Mhm yea

[obsidian marten]

Link for separating lead vocal and back?

[odd shale]

You can either use UVR, UVR Online or MVSEP

What's your gpu?

[restive heath]

i do have rx11

i can do that for u if u want

[smoky linden]

How can i turn off where i cant hear my self

i have monitor on none

and still able to hear me self

[tame mica]

https://tenor.com/view/face-drop-instagram-instagram-live-girl-woman-gif-14331415459530630583

ASSGFUSFHGV

[analog sage]

um hello, how can i use a model that i have downloaded from here?

[low shard]

just login on ngrok

[low shard]

what's ur pc gpu?

[tame mica]

i did

[analog sage]

i am using colab so.. t4. will it work with that

[tame mica]

i put my authtoken on the secrets thing as instructed

idk what went wrong

[low shard]

did u check ur pc gpu first? bc colab is only for people who got a bad pc gpu

[low shard]

maybe try making a new authtoken

[analog sage]

i have not really looked into my pc gpu, i have using colab since the beginning

[tame mica]

that was a refreshed authtoken :sob

[low shard]

it's better u check that

in case ur pc is good enough

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

[low shard]

huh pretty weird, i could just suggest to make a new one

[analog sage]

i checked it, but i am not sure what to conclude from it?

[rare goblet]

Ayo? @analog sage level 1 !!!

[brittle wing]

I used I've De-noise for my dataset but it doesn't remove the noise completely.Someone help please

[azure marsh]

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

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

[brittle wing]

How do I De-noise my dataset completely on Android

[azure marsh]

Dataset

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

[tame mica]

hhh icic

[low shard]

tell me your gpu name, the thing is if its good enough, you can do it locally

[low shard]

did it work?

[tame mica]

no i gave up

ill try tomorrow ig

[analog sage]

i have tried different ways but i can only see that its AMD Radeon.. do you know how can i check the exact or is there no need of that

Radeon graphics one

[low shard]

Like Radeon tm graphics or Radeon Vega graphics?

[analog sage]

tm

[low shard]

Are there multiple like GPU 0 GPU 1?

[analog sage]

only gpu 0

[low shard]

Then your GPU is too weak yea

I had to check bc I seen people with an rtx using colab lol

[low shard]

For that you have to use cloud (remote good pc)

[analog sage]

ok so now you can tell me how can i actually use in lol

[low shard]

You can use either 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! 💖

[analog sage]

use in colsb*

[low shard]

Or applio colab https://docs.ai-hub.wtf/rvc/cloud/applio-colab/

[low shard]

Btw Ilaria rvc zero as it's a ZeroGPU HuggingFace space it has a better GPU (A100) than colab so it will be faster than applio

You can use either one of those tho

[analog sage]

i am gonna try the ilaria one, thanks for the help so far!

[cloud finch]

anyone know what this is?

[rare goblet]

Ayo? @cloud finch level 7 !!!

[cloud finch]

whenever i import my audio its way shorter than it actually is

and gives this error

[low shard]

your're welcome

[rich otter]

Quick question do I need to get a characters high pitch voice aswell? Because when a song does a really high note instead of the character matching it it's stars singing like it's a demon

[worthy quartz]

Anyone know if there’s an option to make your own voice on voiceify ai

[low shard]

voicify.ai is a scam, it uses a free program and makes u pay for it

check my reply #🧬│ai-chat message

[rich otter]

@low shard

[brisk void]

Hello! I was trying to generate a file using Applio but it failed. I have used the model previously so it has to be related to the audio I'm using...I'm getting this error:

An error occurred during audio conversion: cuDNN error: CUDNN_STATUS_NOT_SUPPORTED. This error may appear if you passed in a non-contiguous input.

Does anyone know what does it mean with non-contiguous input? It is a 17-minute wav file which works fine on its own

[low shard]

if you mean adding parts of the character in high pitch voice to the dataset to train

maybe that could help but i'm not sure

[rich otter]

👍

[rare goblet]

Ayo? @rich otter level 1 !!!

[heady mirage]

where is the download link???

[simple ore]

use split audio option

[azure marsh]

Overtraining, or overfitting, in machine learning is when a model learns too much from the training data, including noise or irrelevant details. As a result, it performs well on the training data but struggles to make accurate predictions on new, unseen data.

💡 Tip: An effective way to detect overtraining is checking if the TensorBoard Graph starts rising and never comes back down.

[low shard]

to what

please elaborate more

[brisk void]

It worked, thank you!

[rare goblet]

Ayo? @brisk void level 3 !!!

[brittle wing]

Is there someone who can delete my old posts from my deleted account?

-colab

[azure marsh]

📒 Google Colab Notebooks

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

ℹ️ Note

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

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

[frail umbra]

what program do you guys recommend for text2speech (local) with RVC?

[rare goblet]

Ayo? @frail umbra level 1 !!!

[snow hazel]

Hello. I'm trying to seperate all the harmonies in a song but I don't know how.

There's multiple people singing in the harmony btw

[low shard]

Applio

built-in tts

[azure marsh]

✍ Suggestions

• Search for it in AI HUB Docs or Applio Docs. You will probably find your answer there 📚
• Ask for help in #🔍│help-w-okada if it's related to real time voice changing but make sure to read #1297207135469305866 first
• Ask for help in #✨│ai-help for general help, but use the command !howtoask first to learn how to structure your question properly and increase your chances of getting a reply
• Last but not least, ask for help in #🔍│help-ai-art if it's related to AI images.

[crude mist]

anyone got a recommendation for a female voice model that doesnt sound robotic?

where can i filter by like most downloaded or most epoch

[brittle wing]

For dataset I first extract the Acapella with bs roformer and the backing vocals with Mel roformer karaoke (mvsep or UVRONLINE)?

[tame mica]

both are fine

[odd shale]

Both indeed.

[tame mica]

mvsep has virtually no limit per day but you need to create an account to have faster queues

[brittle wing]

You're confusing

Also do you make models and are you model maker

[odd shale]

That's true.

[tame mica]

though uvronline is way faster (and does not require an account) but has an audio limit of 10 minutes per day

[brittle wing]

Another thing how do you remove that slight thin layer of noise @odd shale

[odd shale]

I guess you can try using UVR Online's denoise.

[brittle wing]

Already did and still got rejected

By the bot

[odd shale]

I'm not really sure to what kind of noise are you referring to sadly.

[brittle wing]

I can dm you my dataset and you tell me all the steps in detail
I'm gonna redo the WHOLE dataset tho

I'll dm you

[odd shale]

what?

I've told you i don't know what kind of noise you're referring to.

[brittle wing]

Tell me the steps for filtering a dataset and what models do you use

[brittle wing]

That's why I have to DM you my dataset and a sample

A slight creaking noise

[odd shale]

Just DM it to see what i can do.

[brittle wing]

Okay and you tell me the correct steps and procedure for a dataset

Cause some people have no idea or are struggling and are inexperienced

It's not easy

[brittle wing]

Uploading

[rare goblet]

Ayo? @brittle wing level 14 !!!

[brittle wing]

@odd shale sent

[azure marsh]

Dataset

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

[brittle wing]

-faq

[odd shale]

Idk what to do to delete that creaking noise sadly.

I'm sorry.

[brittle wing]

Okay tell me the steps you use for making a dataset so I can redo everything

[odd shale]

Have you read the docs?

All i'll suggest is using either latest Mel Roformer or BS Roformer first on MVSEP.

Then Mel Karaoke/and UVR BVE V2 on UVR online.

[brittle wing]

Uh okay send the specific doc

I need help I'm tired of being rejected

[odd shale]

Then use anvuev's Mel dereverb (In case of not working with your audio just use MDX23C Dereverb) and then De-Echo on either MVSEP or UVR Online.

Additionaly Mel Denoise or Aufr's denoise at the end.

[brittle wing]

OKAY but first I remove reverb, then eco?

[odd shale]

Yes.

[brittle wing]

Both are the same model

[odd shale]

I can't suggest anything else besides RX10 edition.

[brittle wing]

I don't have a computer

I don't use*

[crude flame]

Melband duality is also a really good option

[brittle wing]

Also it's paid

[odd shale]

Really?

[crude flame]

Yeah

Just use v1

Not v2

[odd shale]

I'm so sorry then.

[odd shale]

It works for Vtuber Vods and streams?

[crude flame]

#🔥│model-maker-chat message

[crude flame]

Yup

[odd shale]

And for songs?

[crude flame]

Yup

[brittle wing]

What did you say I use for backing vocals extraction
First with bs roformer after that I remove reverb, echo use uvr BVE or Mel roformer karaoke?

[odd shale]

Nice..

[odd shale]

First bs roformer then mel karaoke.

[crude flame]

Sadly it's not on mvsep yet so you have to use it on the music separation colab

[odd shale]

I don't mind that.

[odd shale]

You know, I've realized sometimes BVE V2 can isolate stuff that Mel Karaoke can't isolate. (At least on some songs on my end, correct me if i'm wrong)

So if you want after using Mel Karaoke use BVE V2

That's it.

I can't suggest anything further.

[crude flame]

Also btw don't run audio through too many extraction models because it will start to lose its quality and such

[odd shale]

That's right.

That's another thing to keep in mind.

[brittle wing]

Uh but after using BSRoformer

[odd shale]

I've said Mel Karaoke or BVE V2.

[brittle wing]

I KNOW THAT IT'S SO TRUE.

[odd shale]

Calm down.

[brittle wing]

BSRoformer first right

[brittle wing]

Okay.

[odd shale]

I repeat, yes.

[brittle wing]

Like I don't use Mel Roformer directly for lead vocals I extract Acapella first

Then uvr BVE?After I remove reverb and echo and Mel roformer De-noise?
Did I understand correctly?

[brittle wing]

I wasn't sure

[odd shale]

For last time, BS Roformer - Mel Karaoke - BVE V2 (test if you want) - De Reverb - De Echo.

[brittle wing]

Yey

And for noise?

[crude flame]

Mel karaoke can also sometimes remove reverb in my experience

[brittle wing]

Mel or bv uvr or both one after another?

[odd shale]

Hm, that's a plus then.

[odd shale]

Any of both.

[brittle wing]

Nice but Mel roformer karaoke doesn't remove all backing vocals from the Acapella itself
But from the song/mixture yes

[odd shale]

I don't know how to solve that, since Mel Karaoke works fine for me on both ways.

[brittle wing]

Okay what do I use for noise

Uvr or Denoise by aufr33 aggressive on MVSEP

[odd shale]

If it's for songs use Mel Denoise on X Minus/UVR Online or Aufr33 aggresive on mvsep.

That's all i can suggest.

[brittle wing]

Mel roformer aggressive

I haven't tried still

[odd shale]

Mel Roformer Denoise on X Minus.

[brittle wing]

@odd shale it doesn't remove anything are you sure

Oh wait

But do you use BVE UVr or Mel karaoke before that

[odd shale]

I can't suggest anything else.

[brittle wing]

Uh what's the correct model for backing vocals out of these two

Cause I'm seriously confused

Really

Is that the way you do your datasets

[odd shale]

There's no correct model, it's matter of testing.

So no.

[hollow nymph]

does RVC WebUI not have a text to speech option?

[odd shale]

Also there's no need to ping staffers when you're already talking to them.

[brittle wing]

So if Mel-roformer doesn't remove all backing vocals from the first time I use BVE?

[odd shale]

I guess so.

[brittle wing]

Cause they remain in some songs doesn't work from the first time

[brittle wing]

Alright

I get it

[odd shale]

Idk what's the problem on your end then, i'm sorry.

I've already tried to help.

[brittle wing]

You use the one on UVRONLINE or MVSep

[brittle wing]

I'm grateful

I haven't tried that exact method

I'm gonna try

[odd shale]

UVR Online.

[brittle wing]

Ah same

Thanks

[odd shale]

You're welcome..

[brittle wing]

👍

[crude flame]

Do you have rx?

[odd shale]

He said he doesn't have a PC.

[crude flame]

Oh

Mb

[brittle wing]

No I have audacity web version

[brittle wing]

Helen is my name

I'm not he 😂

It's okay

[odd shale]

Oh my bad, i'm sorry.

🐢

[brittle wing]

Don't you see it from the font and my profile photo...like I toughts it's obvious

[glacial pollen]

@brittle wing I'd keep it nice if I was you.
This is purely the problem of you not including the " pronuns ".
Nowadays people are touchy on that and it's within your interest to make sure it's nicely seen

[brittle wing]

It's okay my name is Elena but I like Helen more

[glacial pollen]

Just a minor note ~

Leo did nothing wrong ¯_(ツ)_/¯

[glacial pollen]

cause it sounds kinda passive aggressive don't you think

[brittle wing]

I don't put pronouns it's unnecessary my name is enough

[glacial pollen]

For some people name isn't enough

you know the deal, too many dramas with " you assume bla bla "

[brittle wing]

I was just suggesting and giving hints how is that passive aggressive?

[glacial pollen]

I ain't from them but heard enough of that yk

[glacial pollen]

well whatever, I just suggested to put it in to avoid future confusion

[odd shale]

Come on Cody, it's fine.

[glacial pollen]

As long you say so yuh, alright

[odd shale]

I don't want fights between you both

[glacial pollen]

not really a fight, was genuinely honest

[brittle wing]

I didn't see an issue?

[glacial pollen]

🥬

[odd shale]

Helen, everything's fine.

[glacial pollen]

^

[brittle wing]

I didn't mean to be rude?

[odd shale]

Nope, you weren't rude.

[glacial pollen]

well that's fine then, I am just a lil too sensitive given I've seen a lot of dramas caused by such things

[brittle wing]

I was just wondering why you got confused since everything's obvious the cursive font the pfp of a girl character

Britney Britney

[glacial pollen]

I mean yeah but you know how it is nowadays

people are people, sometimes net id =/= irl

[crude flame]

Catfishing

[brittle wing]

Uh I prefer not to use pronouns my name is enough.

[glacial pollen]

or that ^ ye

[spring orchid]

a girl on ai hub? no way

[odd shale]

🐢

[glacial pollen]

well then, not to be rude, but you can't get uhhh too mad or uncomfy with people mistaking it
if you decided to not use that informative function

[brittle wing]

Uh I didn't get mad?

I wasn't even mad

[glacial pollen]

but it clearly was a problem to you, no?
if it wasn't a big deal you'd brush it off

[odd shale]

Please don't argue or else tutel is gonna be sad.

[brittle wing]

????

[odd shale]

🐢 🥺

[glacial pollen]

Not arguing tho, yet I get it ye. Gonna go back to my work

[crude flame]

People who catfish do everything you said

[crude flame]

^

[brittle wing]

Do you want me to dm you literal photos of myself

[crude flame]

No

[brittle wing]

I'm a woman bro...

[crude flame]

Ok

Cool?

Want a cookie

[odd shale]

That's alright!

[brittle wing]

No I prefer waffles

[odd shale]

I apologize if Cody got confused.

[brittle wing]

Ikr

[odd shale]

Yep.

[brittle wing]

I was asking about dataset procedure and didn't want this to turn into a fight

[odd shale]

I prefer lettuces.

[brittle wing]

Same with 🍋 juice

[odd shale]

And i decided to stop it because i didn't wanna see Cody discussing with you for something dumb.

[brittle wing]

It's really getting dumb.

I thought so

[odd shale]

🐢

[brittle wing]

I like turtles they're so innocent

[odd shale]

I'm a turtle.

[brittle wing]

A ninja turtle?

Which one

[odd shale]

Nope, a tiny turtle.

[indigo dagger]

any online site where i can upload mp3 and use models and export?

[rare goblet]

Ayo? @indigo dagger level 1 !!!

[paper oyster]

Can anyone help me expand a photos background to be a iPhone wallpaper ?

[rare goblet]

Ayo? @paper oyster level 1 !!!

[brittle wing]

Use ai expand on PicsArt

[brittle wing]

Nice

[devout field]

has anyone got experience with music-gen

i cant make it seem to work

[serene sundial]

is there a good toturial to learn about how to clone a voice with RVC? voice-to-voice conversion

[low shard]

yea, what's ur pc gpu?

[rancid sluice]

hi, where can i find the updated rvc

[low shard]

what's ur pc gpu and are you looking for ai covers or realtime voice changer for calls

[rancid sluice]

realtime voice changer for call

[crude flame]

What the ancient pc

[rancid sluice]

im poor

[low shard]

that gotta be an historical cpu ngl

[low shard]

well, for that there is wokada

[deep linden]

this is literally a server cpu or somethin

[low shard]

especially the fork with better performance

[crude flame]

That needs to be in a museum

[rancid sluice]

yeah

[low shard]

U could try to use https://rentry.co/ForkVoiceChangerGuide as u got an RX 580

[rancid sluice]

bruh

[low shard]

i thought i3 and celeron were the worst cpus

[rancid sluice]

I'm Brazilian, I'm poor XD

[deep linden]

celerons would be literally useless if they didnt have later generation models

[low shard]

Understandable lol

it's just funni as i havent even heard of that cpu before

[low shard]

i thougth celeron was the lowest u could get today tbh

[deep linden]

probably

[low shard]

like the bottom of the barel

i got an i3

which is meh

[deep linden]

i have a laptop with an n4020 celeron processor

its pretty.. unusable

[low shard]

n4020?

[deep linden]

for anything more than minecraft

[low shard]

ohh

[deep linden]

its literally a shock it can even play minecraft

though on that laptop since it has literally like 4 gb of ram you will not be playing singleplayer java

[low shard]

atp make ur own cpu

[deep linden]

bedrock is literally the only way to physically play modern minecraft singleplayer on a configuration that is extremely low-end

[low shard]

bedrock uses less resources?

didn't know that

[deep linden]

there is a video where someone cranked render distance and shit on a pentium

and although there were issues it was actually insanely fast for what it is

he even cranked the render distance

and was still getting like above 60 fps

[rancid sluice]

To this day, I've been able to play everything I wanted on this computer. Elden ring, lol, valorant, rdr2, bg3...

[deep linden]

bedrock is just faster and more optimized than java minecraft will ever be

[crude flame]

Elden Ring?!?! Wow

How many fps

[low shard]

damn

[low shard]

wow

[rancid sluice]

I don't remember, but I think it was in the 60s, because it was smooth

[low shard]

i mean, got the minimum gpu

i think ur cpu kinda bottlenecks it

[crude flame]

he's literally the bare minimum

[low shard]

but hey i thought the requirements were way more demanding tbh

[rancid sluice]

xd

[low shard]

I ran fortnite on my i3 btw

[crude flame]

Huh

[deep linden]

so did i

when i had one

though i had a gt 1030

[low shard]

It was sooooooooo shit

[crude flame]

I don't think I ever had a bad pc

[low shard]

i used it only for UEFN, i was able to get developer shit

[low shard]

lucky rich fr

[deep linden]

imagine how much better that would be on linux since apparently it runs faster on linux lol

ive heard elden ring literally runs better on the steam deck than on windows

[crude flame]

I think my worst PC had a 1660 TI

And that was several years ago

[low shard]

how rich are u

what's ur current gpu

[rancid sluice]

Dude, if my currency wasn't so devalued, and if the government wasn't such a shithole, I'd be in the same situation as you

[crude flame]

6700xt but I could have gotten a 7700xt if I knew that the game that was included with the 6700xt wasn't hardware locked

[low shard]

Yea I heard Brazil currency is fucked up

Btw, we weren't trying to make fun of you, sorry if you understood that

[crude flame]

I am a brokie wdym

[low shard]

mf im on a i3 laptop since 2020

Its begging for death 🙏

[crude flame]

I also have three monitors, two 1080p and my main is 1440p

[crude flame]

💀

[low shard]

how much money do u make being model master

[crude flame]

All 144hz and above

[low shard]

[crude flame]

A big fat zero

I haven't taken a comm

[low shard]

oh damn

u got a job?

[crude flame]

Not yet

[low shard]

then its my skill issue atp

[crude flame]

I spend my bday money so wisely

You don't know

[low shard]

btw if u ever gonna buy a better GPU, i'd suggest for an RTX 3060 12gb, cheaper and still good

[crude flame]

I learnt about PCs for like 2 years before actually getting mine just do I can get a good deal

[low shard]

that reminds me of when i bought once a fn battle pass so it gave me more vbucks than the ones i spent, so i never actually bought any other battle pass
Stonks

not related to bday money

[rancid sluice]

When I'm working I'll do this, for now I'm unemployed, it sucks not getting a good job

[low shard]

but it reminded me of that

[snow hazel]

Hello. I'm trying to seperate all the harmonies in a song but I don't know how.

[low shard]

goodluck

[crude flame]

I did this too

[snow hazel]

Also there’s many people singing in the harmony.

I want everyone separate if possible

[rancid sluice]

thanks

[low shard]

my display is literally 1366x768 😭

obs recordings look awful

[crude flame]

[low shard]

to separate multiple people, u could try to separate the vocals and the instrumentals, then use HP Karaoke 6 on the vocals

not sure if there's a better way

[rancid sluice]

I used this one before, but there probably needs to be a better update now, is there?

[crude flame]

^

[rancid sluice]

same xd

[low shard]

It would be better u use the one I sent you, its the fork that will help u have better performance

I don't think I can do that

[crude flame]

Why?

Do you not have a lawn mower?

[low shard]

iirc there are already adults who do that here, I see some people cutting the grass here sometimes so yeah

[crude flame]

Just do there job for cheaper

[rancid sluice]

I get great performance with RVC, I want performance, but I also want quality, does this offer me both?

[crude flame]

Then bam money

[crude flame]

Or do more for the same price

[low shard]

Yea ofc it offers both, the quality depends by the model tho

[low shard]

mm i will see what i can do

smh im gonna call sleepy joe against u

[crude flame]

You about to be rich fr

[crude flame]

He ain't the president anymore

He can't do anything

[low shard]

smh

[crude flame]

And he will prob forget what you said in like 5 minutes

[deep valve]

guys hmm i have certain weird issue installing VAC, did someone knows how to solve it?? i will explain a bit the issue, when i install the program and reboot my pc, i lose Realtek audio and Realtek inner microphone wich only solves uninstall VAC,

[snow hazel]

On what? X Minis? RVC?

[deep valve]

how i keep the Virtual cable and my normal microphone or what i did wrong

[low shard]

I remember using it locally on UVR, not sure if it's on MVSEP too

[quasi dagger]

You can also try using medleyvox on https://mvsep.com/ but if there are many harmonies you will need to keep rerunning it multiple times to extract each voice but the quality will get worse the more you do it

If it's a simple lead and background vocals the karaoke models are fine

[snow hazel]

MCSEP doesn’t work for follow do UVR will do.

[low shard]

u can also try what raytracer said ofc

[quasi dagger]

That's how I isolate harmonies

[snow hazel]

I looked and I couldn’t find melodyvox on MVSEP

[quasi dagger]

I mean individual notes in the harmony, but for all voices in the background together the karaoke models are best

[snow hazel]

I found it lmao. Hopefully this works.

[quasi dagger]

The quality is very bad though but it's interesting if you want to learn the notes

Also sometimes the voice blend with each other, for instance a tenor line can be on the soprano line etc

[snow hazel]

I don’t know what that means tbh

[shadow grotto]

cani make some voice say something?

[quasi dagger]

But is that what you're trying to achieve or do you want separate the lead from the rest? In that case the karaoke models are the best

[snow hazel]

Tbh I wanted to separate all the vocals and change the voice with RVC.

[quasi dagger]

Ah that's cool, I never tried doing it on the background vocals since the result of Medley is very bad

But you could try

[snow hazel]

I’ve been using X-Minus because it’s been really great in quality but I have to do two things a day cause it locks you out after 10 minutes of audio uploaded.

But I will try MVSEP

[quasi dagger]

In fact people behind X-Minus are the same team of mvsep

But mvsep you can do more for free

Also if you create an account there the queue is not that bad

[snow hazel]

Oh no way! That’s very interesting.

[quasi dagger]

They are always adding new models

[snow hazel]

I see.

[brittle wing]

Can someone help me on how to make a noiseless dataset?

[snow hazel]

Yeah MVSEP didn't work.

[near moon]

what repo are people are using to train rvc models nowadays ive been tapped out

[quasi dagger]

What did it show to you?

That's weird, might be temporarily down

[glad zealot]

I still use mainline

[snow hazel]

Nah it worked but as you said it was just very poor quality.

[knotty moth]

mel roformer karaoke is the best for lead vocals but leaves quiet yet still audible portion of lead vocals in back vocal stem (since it is inverted from the target main vocal stem). UVR BVE may mitigate this issue for back vocals but it's not fullband.

[latent kettle]

I have downloaded the latest version of Applio but when I try to upload index it give me an error massage. And I opened my CMD it says "list index out of range" any fix ?

[idle panther]

Hello, I have a request and I hope you can help me. I used to work with RVC w/ crepe & crepe-tiny (Mangio-RVC-Fork) (EXPERIMENTAL) but now the link is off. I tried to use Ilaria RVC but the result quality seems too different. Anybody know if there's something equal or similar to the old Mangio RVC?

[pastel oak]

Applio uses same crepe as mangio

[pastel oak]

Is there a ran out of input error right after

[pastel oak]

https://docs.ai-hub.wtf/rvc/local/applio/

[latent kettle]

No. Nothing like that

[idle panther]

Nice, I never tried it. What the differences between Applio and Ilaria? Maybe I had no luck or I used Ilaria not properly

[pastel oak]

I suppose index is broken

[latent kettle]

I downloaded it from Weights.gg

[idle panther]

btw thank you!

[rare goblet]

Ayo? @idle panther level 1 !!!

[low shard]

mostly just the interface tbh

[low shard]

could u share model link download

[devout field]

can anyone please help me with music-gen

[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! 💖

[devout field]

ilaria omg can u help me

wait ive been catfished by the app

wtf

[devout field]

sadge

[prime ginkgo]

is there any tutorial of how to train the best rvc model possible for i can read?

[low shard]

What's ur PC GPU

[brittle wing]

Can someone add rmvpe+ in APPLIO colab?
Ilaria tends to decrease volume and quality

[simple ore]

• does not do anything

it simply wipes f0 curve under/over threshold to 0

[brittle wing]

?is it better than regular rmvpe?

[simple ore]

no

[latent kettle]

Every model downloaded from Weights.gg

Other models are working fine

[simple ore]

download a model from weights.gg (comes as a .zip), unzip into a new folder under /logs/

click refresh on Applio inference UI, pick the model

[brittle wing]

Okayyy

[unique rock]

Can someone give me a noise eliminator? I use a mobile phone and I can't find one!

above all, of being.

[brittle wing]

https://github.com/camenduru/Retrieval-based-Voice-Conversion-WebUI-colab
Are these up to date?
Uh wait they're...

https://voice-models.com/colab

Y'all so many colabs exist.

[pastel oak]

These are the best options (ripped off from Nicks copy paste)

training models:

• 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)

[brittle wing]

Okay but where's the newest ai cover generator

[pastel oak]

i recommend ilarias space

[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! 💖

[brittle wing]

No it generates noise

[pastel oak]

doubt that but sure, then try applio colab
https://docs.ai-hub.wtf/rvc/cloud/applio-colab/

[brittle wing]

I have tried both I don't have problems w applio but what I'm asking is where's the newest slow ai cover generator

[pastel oak]

If there is a new one i dont know about it

[brittle wing]

Uh ok

[karmic kite]

does anyone know what RVC collab is even used anymore? the old one is gone and I haven't touched this shit in like a year

woah I'm green

but uh yeah I've been using weights for ages now and

the voice I wanna use isn't on it

[last portal]

-svc

[azure marsh]

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

[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! 💖

[brittle wing]

Are 30 minutes enough?

[crude flame]

Yuh

[brittle wing]

Should I go further?

[crude flame]

If you want

[brittle wing]

I know how to make a dataset now someone told me a meithod that works

[brittle wing]

Is it necessary for a better model

[crude flame]

Yup

More data = more natural sounding

[brittle wing]

And more steps/epochs

[crude flame]

More epochs doesn't mean better model

Steps is just how many times the models weights are being adjusted

[brittle wing]

Ikr what I care about are the steps now

[crude flame]

More steps also doesn't mean better model

[brittle wing]

Yeah but with Titan pretrain it reaches a huge amount of steps even on a small amount of epochs

[crude flame]

To get more steps you just decrease batch size or get a longer dataset

Doesn't have anything to do with pretrains

[brittle wing]

Huh

I didn't know my previous one was about 90k steps with 8 batch size 30mins data

[crude flame]

With how many epochs

[brittle wing]

200

[crude flame]

Sounds about right

[brittle wing]

Uh really

[crude flame]

30 minutes of data 200 epochs 90k steps sounds about right

[brittle wing]

How do you know that

[crude flame]

Experience

if you want to calculate the steps for a specific epoch you would use this:
Steps = dataset length / batch size * epochs

If you wanted to know

[simple ore]

30 min = 1200 3-sec slices

1200 slices / batch 4 = 300 steps / epoch

300 epoch = 90k steps total

[brittle wing]

Kinda hard to understand and mine was over 90k at 200

[simple ore]

smaller slices

[hasty umbra]

hey can someone assist me with an rvc issue? a vol value goes up but for some reason it doesn't process the output that i can hear

im trying to route a different software input into the RVC, and i do see a volume input based on the stats it shows, but i cant seem to hear its output

[simple ore]

RVC != realtime voice changer

[hasty umbra]

well its using RVC tech

[rare goblet]

Ayo? @hasty umbra level 4 !!!

[hasty umbra]

but it wont process the audio for some reason i cant figure out why

its a super niche issue with an already niche program, so im struggling to get any answers at all

[simple ore]

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

[hasty umbra]

ah ok

[sacred monolith]

colab link?

[junior hawk]

can i add one of the voices to tts voice wizard?

[low shard]

To what?

Elaborate

And colab is a cloud computing service only for people who got a bad PC

Tell ur PC GPU first

[low shard]

Don't think so,
RVC is made for Speech To Speech

Not for TTS

[low shard]

There's sooooo many people thinking RVC means that and using this channel lmao

[simple ore]

how about renaming help-w-okada.. cuz wtf is okada?

[low shard]

I mean everyone calls it Wokada

We can't just put realtime voice changer bc we don't use other garbage ones like voice.ai

[pastel oak]

Used to be called help voice changer but people still asked in here

These days people dont know its called wokada btw @low shard the amount of times i say "send ss of wokada" and they ask whats wokada

[low shard]

Skill issue ngl

It's like how people still asked how to make models in be model maker

That rename helped for a while but then I still seen people asking there

[pastel oak]

Follow this step:
https://rentry.co/ForkVoiceChangerGuide#press-enter-to-continue-failed-to-download-or-verify-files

section linked in the url

[low shard]

Every guide calls it wokada

Except yt tuts

Voice changer is too generic and people don't like reading

[pastel oak]

for a while people called it the "mmvcserversio", but now wok changed the labelling again to vcclient, so we are going down another path

[low shard]

Smhh

Every of our guide calls it wokada