✨│ai-help

[tranquil cliff]

thx

[flint vapor]

ti mando anche un screen di come è l'interfaccia

[flint vapor]

poi per scaricarlo in locale da qua https://github.com/Mangio621/Mangio-RVC-Fork/releases/download/v23.7.0/INSTALL_Mangio-RVC-v23.7.0_INFER_TRAIN.bat 😄

[flint vapor]

aspe te la mando in dm l'interfaccia che si vede il mio nome

[tranquil cliff]

okkk

[flint vapor]

Tell me if you resolved after 😄

[young halo]

Ohhh, can I fix that?

[tranquil cliff]

just start generate index first

[young halo]

Can I do so if my model is already training

[tranquil cliff]

no

[young halo]

Oh, so i have to stop it then

[tranquil cliff]

yes

[rare thorn]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[night relic]

Does any one know how to download large files from MEGA in less time and without Transfer qouta issue

[feral saffron]

-Colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[odd shale]

I think you could use Megadownloader.

I did use it a lot years ago when i barely had an potato laptop.

[night relic]

😂🤣

[inland sonnet]

Hi could someone help me with the voice changer client demo plz ?

[charred topaz]

Im having the same problem in both WEB UI and Mangio, the first voice model i created in those 2 programs worked really well, but after those, now the traning is way too fast, like 1 second per epoch, and the result is really bad, the first time in both of the programs each epoch needed like +1 min

[weary delta]

how to use the voice model that I downloaded if I want to voice my text?

[worthy tulip]

@sand iris do you know if there are any working rvc inference's for colab? im on mac n the infer i usually use has stopped working

if not i might have to try installing applio thru terminal

[sand iris]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[sand iris]

Ilaria RVC

Stuff is kinda broken on mac, I'm not sure what would work

@worthy tulip

[worthy tulip]

i'll try it out thank you!

[rare goblet]

Ayo? @worthy tulip level 1 !!!

[odd shale]

I think Ilaria RVC has TTS.

So you can use it

[spiral osprey]

I can't train my voice model in RVC Disconnected V2

someone help-me pls.

[olive crow]

does somebody know if i have overtrained my model?

[proper shale]

yeah no not overtraining yet

[olive crow]

how many models have you trained?

[proper shale]

like... over 10 at this point

[olive crow]

nice

[sly pilot]

hello, I've been trying to train a voice in this website "https://colab.research.google.com/drive/1TU-kkQWVf-PLO_hSa2QCMZS1XF5xVHqs?usp=sharing&authuser=5#scrollTo=7vh6vphDwO0b" but I'm getting an error "AttributeError: module "huggingface_hub.constants' has no attribute 'HF_HUB_CACHE" whenever I try to do the step 4 by running the gui

[dusty rampart]

not overtrained but seems to have a lot of collapses

turn off the smoothing rq

[dusty rampart]

you're using an old and discontinued colab

use one of these

-rvc

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[spiral osprey]

pls help-me

[dawn nebula]

did u run feature extraction?

[spiral osprey]

yes.

[dawn nebula]

run it again ig?

[spiral osprey]

ok

[olive crow]

Why?

[dusty rampart]

to see collapses

[olive crow]

Ok

[low shard]

Did it worked? Or do you see "no feature todo" at feature extraction cell

[spiral osprey]

again

[spiral osprey]

@low shard again lol

[dawn nebula]

try disconnect and delete runtime

and run it again

[low shard]

Show me the output of the feature extraction cell

[spiral osprey]

I've tried it before and it does the same thing.

[low shard]

Its prolly the no.feature todo thing

[proper shale]

uhhhh not necessarily

[proper shale]

ya screwed up on the dataset formatting OR your audio screwed up

Make sure that:

1. dataset folder name is the same as the experiment_name in the Set Training Variables step.

2. you're not compressing the files as .rar. this causes issues.

3. in case you did all of those, it could probably be an audio issue, try a re-export in Audacity as a .wav with all audios you used.

[brittle wing]

Hey is anyone here to help me out quickly ? D:

[spiral osprey]

None of these are my problem.

[proper shale]

State the issue and we'll try

[proper shale]

Just make sure your audio is not screwed up, and then the index training will work. That's what I wrote there basically

[brittle wing]

Simply put.

I followed the Instructions from this website https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=Y13Eh9r_g8f-

But whenever i do the "Download The Model" i get a an error message

---

JSONDecodeError Traceback (most recent call last)
<ipython-input-15-cd6eef5bf3c2> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:

5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end

JSONDecodeError: Expecting value: line 1 column 1 (char 0)

I don't know, if thats an issue on my part and I have to change something.

[rare goblet]

Ayo? @brittle wing level 1 !!!

[proper shale]

Google Drive links are broken... if that's what you're using

Oh and that Colab is outdated

[amber thorn]

I'd assume its colab's being outdated fault, since i have the same issue

[thin stump]

Sorry for the ping @fluid matrix. I saw yours and Lusbert's recommendation for Renegate. Do you still have your preset from this outdated guide of yours https://rentry.org/HowToRVC?

[amber thorn]

tried multiple different links but didn't work

[dawn nebula]

outdated colab, use rvc disconnected

[proper shale]

yeah tbh same

[brittle wing]

I see I see

[proper shale]

what if they wanna do covers

[dawn nebula]

aicovergen

lol

or illaria rvc

[amber thorn]

cheers lads

[clear lily]

when ever i upload a ai voice in RTVC i get this error

any fix?

im really new to this stuff

[proper shale]

you're uploading the D pth (D and G are for training) and the trained index (the added index is the one you're supposed to use)

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[clear lily]

I'm so dump

[rare goblet]

Ayo? @clear lily level 3 !!!

[clear lily]

Thanks

[proper shale]

Nah, you just gotta get used to stuff :)

[proper shale]

You're welcome :)

[odd shale]

These kind of errors are my big weakness.

[brittle wing]

is there a way to use tts in the rvc so u can basically use tts as ur mic

like mangio rvc

[odd shale]

Ilaria RVC mate.

People wonder why the D and G filenames tend to have the 233333 name but they commit the mistake of having both "save only final epoch" and "save epoch each checkpoint" options marked.

[brittle wing]

its a website i cant download it

[spiral osprey]

I'll try with another dataset tomorrow, if it doesn't work, I'll remake the dataset again.

[odd shale]

No, it's an Colab.

[olive crow]

How do i access a check point

[odd shale]

What do you mean?

[brittle wing]

well either way idk how to use it as a microphone output on discord

[sand iris]

old deprecated & broken kalo colab

[odd shale]

Well, it used to be the most popular colab for RVC usage

[brittle wing]

@odd shale where can i find resources to use ilaria as tts microphone

[rare goblet]

Ayo? @brittle wing level 1 !!!

[odd shale]

💀

Ilaria RVC has TTS

[twin heron]

That shit don't work on Windows or Chrome either

[brittle wing]

yea but how do i get it and use it as an audio output on discord

[odd shale]

Sorry but i don't really understand.

Oh, do you wanna use a model on realtime?

[brittle wing]

yes

real time tts

when i cant actually talk at the time and my friend cant check the text

[odd shale]

Sorry but i think there's no tool for what you want.

The nearest thing is W-Okada, with that program you can use models on realtime.

[brittle wing]

i can use the vb audio cable thing too

[odd shale]

Yes you can

[brittle wing]

can i aalso use mangio rvc for text

I have w okada i think

[desert niche]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[glad veldt]

can someone help me with cleaning up an audio file?

specifically this

[rare goblet]

Ayo? @glad veldt level 3 !!!

[thin stump]

Here's a workflow

• MVSEP (MDX-23C)
• UVR BVE model, or HP_5/HP_6 to isolate lead or backing vocals (your song doesn't have any)
• Melband Reformer on MVSEP
• UVR-DeEcho-Dereverb 0.3 aggression
• UVR-Denoise 0.5 aggression

[brisk dew]

what can i use the index file for?

[latent kettle]

installing RVC on mac getting errors please help

[latent kettle]

it stores the accent of your voice model

[young halo]

I tried to train the index of my model on Applio but it appears like this...

And the tensorboard doesn't show anything

[brisk dew]

oh ok

[latent kettle]

it is not index

[latent kettle]

can someone please help. I'm trying to install RVC on mac and getting bunch of errors

[glad veldt]

alright got this

is this acceptable for getting put in a dataset?

im still hearing some reverb and echo

[young halo]

I mean, when I clicked train index that showed up on logs folder

[latent kettle]

it is good

[latent kettle]

index is a log file

[young halo]

Really? Cause its size Is zero

And that's not how it looks like on the guide

[latent kettle]

no .. size is not ok

you have to train it again

[brisk oak]

whats best egirl voice atm?

[thin stump]

Nope, you still want to remove those so you have to toy around with the dereverb-deecho aggression

[young halo]

I did so and it appears the same 🥹

I trained a index for that same model o RVC Disconnected, I don't know if that works

[thin stump]

normalizing on audacity will come into play later but it won't remove reverb or echoes

[young halo]

And I don't know why there's not a single graph on my Tensorboard

[young halo]

@molten pecan can u help me out?

[tacit grail]

quick question, whats that site for isolating vocals and instrumentals again?

is that better or UVR?

ah ok. i kept getting the wrong results when i tried searching it lol

yea i got it thanks

does that or UVR get better results?

oh ok

yea i prefer uvr but i much prefer being able to reliably be able to watch stuff while waiting without my pc lagging a bunch while rendering locally

usually takes 2 and half hours on average for each render for some reason on uvr

yea ive used it before but i never could tell if the results were better or worse than uvr so i havent used it mch but if its the same program then ill just use mvsep

when you say that is that literally or metaphorically? like is it made by the same person or is it just the fact that it does the same thing just different quality results

[molten pecan]

Working on the docs. I can't

Here's a small peak for what's coming

https://rentry.org/colab_workarounds

[primal vector]

Keep getting this error when starting Index Training

[tacit grail]

Oh, ok. Well im wondering which service has the best results overall quality wise

if theyre the same then ill use mvsep but if uvr has better quality results then ill have to use uvr because i wanna keep making song covers but at the best quality i can

[glad veldt]

alright got this after using 1.0 aggression

[young halo]

Oooh I understand

[tacit grail]

Alright thanks

[thin stump]

0:50 is suss. I think it's good enough though?

[tidal cairn]

Is there any RVC for inference currently working I went down the list and they all are stuck processing

Bro someone please help

[glad veldt]

I mean if it isn't good enough what more can I do?

Literally used max aggression to clean it

[young halo]

ಥ⁠‿⁠ಥ

[glad veldt]

I think I'd have to manually edit out the remaining reverbs and echos (which is something I don't know how to do whatsoever) if it's even possible.

[tidal cairn]

I remember when getting problems resolved in the server was a priority. Now I just get ignored 😂

[tacit grail]

On MVSEP whats the difference from "Extracting directly from mixture" and "Extract from vocals part"?

[forest sluice]

Applio doesn't work

?

And your problem is not at all a trivial problem

All the stuffs that work are included. (If it lacks anything, make a suggestion)

[icy marten]

Getting an error when I try to use mangio rvc

[rare goblet]

Ayo? @icy marten level 1 !!!

[icy marten]

2024-02-26 23:54:23 | ERROR | root | Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 80, in _run
  File "asyncio\proactor_events.py", line 162, in _call_connection_lost

If anyone knows where this is coming from please let me know ):

[forest sluice]

Directly from mixture: extract straight from the original music
From vocals part: from the instrumentals

[tidal cairn]

it's not working its just on an infinite processing loop

[brittle wing]

hey guys, I'd like to train again a RVC voice, is it interesting to keep background vocals to have a better result with songs that's gonna also have background vocals?

[tidal cairn]

infinite time loop processing bruh

[tacit grail]

what about SDR? is a higher SDR better quality or lower? what does it even mean?

[glad zealot]

It would make it sound worse

[brittle wing]

aight

[glad zealot]

Remove all background as much as possible

[tidal cairn]

no one knows out of 300K people what could be wrong with the RVC inference processing times?

[glad zealot]

Gpu maybe

[tidal cairn]

It’s been working completely fine and fast until today though

[glad zealot]

Are there any error in the console?

[forest sluice]

Inference right? Mine did a TTS in Ilaria RVC and it works

just fine

[brittle wing]

I don't see anywhere on the tutorial website, is it still requiered to slice audios?

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[celest relic]

its errored out everywhere lol

[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.

[thin stump]

You're talking about audio labeling. It's more precise than truncate silences but you have to completely clean your dataset first

https://aihubdocs.github.io/en/vocal-isolation--datasets/datasets/#section-4

ctrl+f for audio labeling https://rentry.org/RVC-dataset-RX

[merry spindle]

does this mean drive or my computer

[olive crow]

been leting it go for 12 hours

overtrained?

[merry spindle]

errrm

[sand iris]

Google drive

there's a file explorer on the left tab of the colab page

[merry spindle]

do i put "my drive" in place of content

[sand iris]

⚠️

[tacit grail]

?

[sand iris]

content from possibly copyrighted audio detected

[merry spindle]

didnt work

do i just make it content

[tacit grail]

its a tiny piece of the vocals

[olive crow]

do you guys know if my thing is overtrained or not

[sand iris]

to the left you have this

[thin stump]

You're rooting for it to go down even further. Idk keep going

[merry spindle]

its in my drive

what else do you want from me

[sand iris]

find your dataset folder (that you must upload to google drive)

[olive crow]

ight

[sand iris]

right click on the folder containing all the audio files, and copy path

[tacit grail]

so no help what so ever just because its from an actual song?

[rare goblet]

Ayo? @tacit grail level 8 !!!

[merry spindle]

is this the best method

[sand iris]

I mean, the issue is with sharing the samples that could possibly get the server nuked

[sand iris]

yes

[tacit grail]

i can understand but its just a tiny piece of the vocals

[thin stump]

if it stops going down and never reaches a new low then that would be time to stop

[olive crow]

ok thanks

[sand iris]

Copyright holders are tyrants unfortunately

and discord just doesn't care and dumbly just proceeds with whatever they're being asked for

[sand iris]

what was the issue with the vocals ? I don't have any speakers to listen to it right now ?

[merry spindle]

also is there gonan be a link to the pth and index of my voice after i finish training?

[sand iris]

it should save in your drive

[olive crow]

i put it it to like 1000 epochs and like 2 hours of clean voice audio for it to train on

[merry spindle]

alright

[tacit grail]

I was just trying to show exactly what i mean for easier understanding of the situation but ok...

Its completely mispronouncing the word "world" even after the vocals are cleaned up

[merry spindle]

world

[sand iris]

what's the "feature retrieval rate" set to ?

or "index ratio"

or whatever it's called ?

[tacit grail]

search feature ratio?

[sand iris]

For real time, the higher values tend to mispronounce, but that probably doesn't transfer to RVC when it's used for file conversion ?

[sand iris]

yeah

[tacit grail]

0.75

[merry spindle]

also is 500 epochs good

for a like

1 minute audio

[sand iris]

I guess you could try lower just for testing ?

I've also heard that after there's a lot of silence, RVC tends to break. Also has issues with pronounciation at the very start of the audio

[merry spindle]

will rvc ever become money hungry

[sand iris]

wdym ?

Online platforms can do whatever the hell they feel like

[merry spindle]

like

[sand iris]

but as the RVC code is open, you always have the option to use it on your own machine

[olive crow]

yo do you guys know where stop training is

[merry spindle]

paying a subscription service to use ai voices is a bit messed up

[tacit grail]

yea ive tried all the way from 0 and 1 and everywhere in between and it still does the same thing

[merry spindle]

is tthis going be an issue

[sand iris]

yes

[olive crow]

yo do you guys know where stop training is

[merry spindle]

how

[tacit grail]

#✨│ai-help message Forgot to click reply

[thin stump]

The button where you started training is the stop

[sand iris]

ah

[thin stump]

Then your folder on the left has your weights file with the step

[olive crow]

but its just says

[rare goblet]

Ayo? @olive crow level 4 !!!

[thin stump]

Oh i thought you were using colab nvm

[olive crow]

no im using rvc webui

[tropic egret]

is there a way to train voice models on an AMD graphics card with directml?

[sand iris]

Unfortunately no

Sadge

All the RVC forks also dropped support for DirectML inference

[tropic egret]

damn nvidia also made it illegal for amd to use zluda's project cuz that would be dope for ai training

[sand iris]

Did they ?

[tropic egret]

ye

[sand iris]

Should be in the clear technically

[merry spindle]

how is drive being uerm
silly

gonna be an issue

[sand iris]

but zluda never really worked on torch

[sand iris]

if it's full you can't save files to it

[merry spindle]

okay no ai for me

:[

[tropic egret]

i think amd was about to work on that issue, the main goal was to be able to train ai models using the cuda's technology

[sand iris]

I mean

Look forward to this :

https://github.com/pytorch/pytorch/issues/106608

[merry spindle]

do i jjust close the tab or will it magically work again

[sand iris]

...But as far as GPU support goes

[fringe zealot]

keep getting ModuleNotFoundError while using the google form version of rvc v2

[sand iris]

it will be something along the lines of :

• vega (56, 64 & radeon VII) or not, if they drop support
• rx 6800+ (for rx 6k cards)
• rx 7xxx cards

[olive crow]

Its been 12 hours and it has 289 epochs, im actually gonna be here forever

[sand iris]

so not great unless you have the shiny new cards

[sand iris]

Which one are you using ?

[fringe zealot]

which what?

[sand iris]

there's many google colabs for RVC

[tropic egret]

my gpu luckily is supported but i wonder if it would work on a radeon 610m (ryzen 5 7520u igpu) since they're both RDNA2

[rare goblet]

Ayo? @tropic egret level 1 !!!

[fringe zealot]

its called Easy GUI (for RVC v2, with crepe) (with improved downloader)

[sand iris]

Oh I think that's the old one that won't work

[sand iris]

I don't think it is

I'm on a 680m on this laptop (6800HS's iGPU), but on linux, and while GPU support is the exact same

[tropic egret]

even if it works it would take a batshit ton of time to train them

[fringe zealot]

whats the new one called?

[woeful cave]

is there a fixed version of the rvc autosave colab..

[sand iris]

it doesn't work out of the box and I have to spoof the card, which apparently isn't available for the windows equivalent

[sand iris]

Are you training a model or doing voice conversion ?

[fringe zealot]

voice conversion

[rare goblet]

Ayo? @fringe zealot level 1 !!!

[sand iris]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[sand iris]

You're looking for Ilaria RVC then

[fringe zealot]

alright thanks

[modern gust]

hi guys, do you know if they are a amd gpu supported version for rvc ?

[dawn nebula]

you have to use linux to train

[modern gust]

sad, i though with the release of zluda, they could have a supported version

[sand iris]

Nah

ain't good enough for pytorch

[tropic egret]

@sand iris have you ever used python venv on linux?

[sand iris]

Yes

[tropic egret]

and

[sand iris]

Just works™️

As long as you don't move it after creation

[tropic egret]

have you ever tried to install beefy packages and get an error that says you don't have enought space

[modern gust]

i test a zluda supported stable diffusion version, it work really good

[rare goblet]

Ayo? @modern gust level 1 !!!

[sand iris]

Oh interesting ? Is that on a custom backend or is it still on torch ?

[sand iris]

Nope !

[modern gust]

no torch

[sand iris]

Though the worst I've done was installing the ROCm torch builds which are like 3Gigs or so ?

[brittle wing]

can soemoen help me

like my shit

isnt working

like everything works app but i cant hear my self nor the voice actully work

[tropic egret]

i've got something like this

[modern gust]

window supported zluda

[sand iris]

Is the venv on another drive ?

[tropic egret]

main nvme

[sand iris]

ah fuck

[tropic egret]

and does it create a virtual partition?

[sand iris]

so ~/.cache would be on the same drive ?

[tropic egret]

i guess so

i only have one partition on this system

except the zram boot and ykyk

[sand iris]

Oh wait I do remember something like this happening ? I'm not so sure

From a quick online lookup, apparently it downloads to /tmp ? that seems silly

[upper furnace]

Help resuming modell

[tropic egret]

yh ur right

[forest sluice]

Show me the output of df -h?

[tropic egret]

it makes a temporary file system as i can see but idk how to resize it if there's a limit (?)

[tropic egret]

is the IFree column shown as bytes or what?

[rare goblet]

Ayo? @tropic egret level 2 !!!

[sand iris]

inodes, so irrelevant

df -h if you want human-readable output in volume

[tropic egret]

oh

didn't realize about that option

apparently tmpfs is 3.5GB on available space

/dev/nvme0n1p2 468G 120G 326G 27% /
tmpfs 3.6G 13M 3.5G 1% /dev/shm
tmpfs 3.6G 9.9M 3.5G 1% /tmp
/dev/nvme0n1p1 511M 82M 430M 16% /boot
tmpfs 718M 140K 717M 1% /run/user/1000

[stiff goblet]

-audio

[azure marsh]

📚 Audio Guides

• Perfecting Audio Isolation on Low-End Rigs: A Practical Guide, by Litsa The Dancer and Faze Masta Google Docs
• Gathering and Isolating Audio, by SCRFilms Google Docs
• Creating Datasets for RVC using iZotope RX, by LUSBERT Rentry
• Vocal Mixing Tutorial, by Roomie YouTube

🛠️ Tools

Audio Separation/Isolation

• https://mvsep.com/
• https://multimedia.easeus.com/vocal-remover/
• MDX-NET Colab

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

[sly pilot]

hello, I'm using Rvc disconnected but Idk how can I resume epoch training after for example restarting the website

[rare goblet]

Ayo? @sly pilot level 1 !!!

[odd shale]

With website you mean colab?

Which colab you used?

[sly pilot]

this one https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS?authuser=5#scrollTo=Yj-_npuA_HRB

[proper shale]

Same steps as if you would train a new model, but with the same training variables. Skip preprocessing and feature extraction, index training too, then just use the two steps for resuming and train with the same batch size as before

[sly pilot]

do I also skip Load Dataset and Save preprocessed dataset files to Google Drive?

[odd shale]

Idk if it's me but i can't access the notebook bc it seems to have been deleted.

[odd shale]

Yes, you should.

[glad zealot]

its rvc v2

disconencted

[odd shale]

Im on phone, so it's pretty weird.

Maybe it's some bug on mobile Google.

[glad zealot]

remove the ?authuser=5 on the link

[sly pilot]

that's weird, when I try to Load preprocessed dataset files from Google Drive it says that it cannot find or open /content/drive/MyDrive/rvcDisconnected/.../rvcLogs.zip

[proper shale]

Yeah

[olive crow]

update on the thingy thing

[proper shale]

Did ya set the same name of the model before and are you on the same Google account?

[proper shale]

solid graph, what about g/mel?

[proper shale]

oh there's that

[olive crow]

where is that

[proper shale]

in scalars tab too, just search

[olive crow]

this

[proper shale]

looks pretty good

[olive crow]

ran it last night its at 336 epochs

running it until it reaches 1k

or 500

[brittle wing]

Is this good?:
https://colab.research.google.com/drive/1PorBmm4bUjY1-dk4f7XKrO7Mx51wbqYU

[rare goblet]

Ayo? @brittle wing level 5 !!!

[sand iris]

I think that's the righ one, yes

or wait

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[sand iris]

https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw

newer version ^

[sly pilot]

I did, btw the reason I was disconected was because I can't use a gpu after an amout of time so I just get some errors when reconnecting without a gpu

[brittle wing]

i started traning on older one. next time

@sand iris how to get model maker tag?

[flint vapor]

You should upload the model on #1159514067187277865

and wait that a model qc will test it

and if all is ok you will obtain the role

[brittle wing]

ok

[brittle wing]

how to upload?

[flint vapor]

https://aihubdocs.github.io/en/extra/model-maker-role/

here is the guide 😄

[brittle wing]

thanks

[flint vapor]

Yw 😄

[brittle wing]

colab disconnected saying that i am running a code which is not the part of free tier

@flint vapor @sand iris

[mortal mantle]

hey

[brittle wing]

i was just able to train 60 epochs

but the inferred audio sound similar to the person's voice. JUST 60 epochs!

[sand iris]

Yes

the tensorboard is a lie !

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[proper shale]

Oh then I suggest waiting another 24 hours and seeing what ya can do by then

just be careful

and limit the training sessions

[sly pilot]

I'm making 1250 epochs, how do I limit my sessions

[restive badger]

what is this ?
JSONDecodeError Traceback (most recent call last)
<ipython-input-5-1d7b23896ae4> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:

5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end

JSONDecodeError: Expecting value: line 1 column 1 (char 0)

[proper shale]

like, stop training after an hour

or two

then export model and come back later

[sly pilot]

alright

[proper shale]

prolly using an old colab

[sly pilot]

how long can be later

[restive badger]

how do i find the new one

[proper shale]

like tomorrow, at least 24 hours id say

[proper shale]

https://aihubdocs.github.io/en

has everything you need

[restive badger]

thank you

[rare goblet]

Ayo? @restive badger level 1 !!!

[proper shale]

You're welcome! Let us know if something goes wrong. :)

[tacit grail]

why cant i run applio? it says "cannot find empty port" when i run it

[cold sparrow]

Does having a diverse dataset of audio like (wispering,talking normaly,talking loud) is batter for a model or unnecessary

[flint vapor]

Should talk normal, but if you want a singer is recommended also dataset of him/her singing

remember that they should be clean very well

[cold sparrow]

Yeah got it thanks a lot

[flint vapor]

You welcome 😄

[cold sparrow]

Also one question about legality is making models on paywalled content "Legal" without publicizing it

[tame lagoon]

I can't load the voice model, it gives an error

[brittle wing]

the tensorboard graph doesnt show/

[brittle wing]

... why? i was using :

https://colab.research.google.com/drive/1PorBmm4bUjY1-dk4f7XKrO7Mx51wbqYU

@sand iris

and how do i resume training from 60 epochs?

[mint delta]

what does the index bar do

[sand iris]

Uh did it save the _G and _D model files ?

[sand iris]

dated

[sand iris]

changes how creative the model can be

at max (1.0) it only produces exact sounds from the original voice of the model

at 0.0 it tries to guess how it sounds

any intermediary values do a mix of both : exact source sound + model guessing

[mint delta]

so 0.5 is good?

[rare goblet]

Ayo? @mint delta level 1 !!!

[sand iris]

Try comparing between the values honestly

[mint delta]

is there also any way for me to disable voice playback?

[sand iris]

wait you are doing real time voice changing ?

keep index at 0 then, it causes performance issues

[mint delta]

okay thank you

[sand iris]

https://rentry.co/VoiceChangerGuide
Almost all the information you could need is found in this guide

but if you need anything feel free to ask in #🔍│help-w-okada

[brittle wing]

yeah

[sand iris]

oh perfect

Not sure how this goes usually, but it's just a matter of using those as "pretrains" if it lets you

[brittle wing]

and whats this?

[sand iris]

I don't do colab model training though

[brittle wing]

i also dont want to. i just have no other option

[sand iris]

When training RVC models, you start from "pretrains", which are generic voice models.
There's multiple pretrains out there, and OV2 is one of them

[brittle wing]

should i tick?

[sand iris]

if you continue from your 60 epoch _D and _G files it won't matter

[brittle wing]

i am not continuing

[sand iris]

oh

then give it a shot I guess ? Not sure what people think of OV2

[brittle wing]

bec i dont know how to continue 🙂

[brittle wing]

u have a guide on how to continue?

[sand iris]

¯_(ツ)_/¯

[brittle wing]

xD

[glad zealot]

as long as you have the log files it should continue

[brittle wing]

how to?

[sand iris]

we're migrating from another session

on an older notebook

[brittle wing]

new google account too

[glad zealot]

fck

is it saved on drive?

[sand iris]

_D and _G are saved apparently

[brittle wing]

yeah

[sand iris]

so should be good, just needs to be able to conveniently use those

[glad zealot]

i think it needs the whole logs folder

[brittle wing]

https://drive.google.com/drive/folders/1Fa2yob0LyfJHkMROg6TFx2DFoTc4Iod4?usp=drive_link

[glad zealot]

oh wait no

can just redo everything

it should resume from the G

as long as the name of the model is the same as the folder in logs

[glad zealot]

oh you have everything

you can resume no problem then

you need to download that and reupload it on the other account tho

since i dont think theres a way to directly move/share it that colab can read diretly

[brittle wing]

how to reupload and continue? is there any guide?

[brittle wing]

hmm

[glad zealot]

i dont think there is

you basically just need to put that to the logs folder and use the same name as the folder on the name input

[brittle wing]

then how can i resume from a new session?

[glad zealot]

download and reupload that logs file on the drive on the other account

[brittle wing]

i need to run everything as it is?

[brittle wing]

feature extraction and everything else

[glad zealot]

basically what you setting you have before you put that again

you dont need to run feature extract and preprocess

when resuming

[brittle wing]

ok

so i run this:
Install to Google Drive (for Resuming Training & Automatic Saving)

then this:

[brittle wing]

when will i put the logs folder and where?

[glad zealot]

the model_name needs to be the same as the name of the folder of you log files

[brittle wing]

ok

[glad zealot]

before you press the train model

[brittle wing]

so i replace the logs file with new one

? @glad zealot

[rare goblet]

Ayo? @brittle wing level 6 !!!

[glad zealot]

put it insinde the logs folder

[brittle wing]

logs inside logs?

[glad zealot]

it should be logs/<model_name>

[brittle wing]

ok. what if i need to continue training from same google account?

[glad zealot]

same thing?

i dont really use disconnected so idk how it works with saving stuff

[brittle wing]

but then inside logs, my model folder would already be present

[glad zealot]

ye then no need to do anything

just start training

make sure the name is the same

[brittle wing]

ok

done

[glad zealot]

make sure all the settings are the same as before

[brittle wing]

thanks

ok

THE END

[glad zealot]

except for the epoch count, you can set that to whatever you want

[glad zealot]

headpat~1

[brittle wing]

ok thanks

from tensorflow.python.data.experimental import service
File "/usr/local/lib/python3.10/dist-packages/tensorflow/python/data/experimental/service/init.py", line 419, in <module>
from tensorflow.python.data.experimental.ops.data_service_ops import distribute
File "/usr/local/lib/python3.10/dist-packages/tensorflow/python/data/experimental/ops/data_service_ops.py", line 25, in <module>
from tensorflow.python.data.ops import dataset_ops
File "/usr/local/lib/python3.10/dist-packages/tensorflow/python/data/ops/dataset_ops.py", line 33, in <module>
from tensorflow.python.data.ops import iterator_ops
File "/usr/local/lib/python3.10/dist-packages/tensorflow/python/data/ops/iterator_ops.py", line 41, in <module>
from tensorflow.python.training.saver import BaseSaverBuilder
File "/usr/local/lib/python3.10/dist-packages/tensorflow/python/training/saver.py", line 50, in <module>
from tensorflow.python.training import py_checkpoint_reader
File "/usr/local/lib/python3.10/dist-packages/tensorflow/python/training/py_checkpoint_reader.py", line 19, in <module>
from tensorflow.python.util._pywrap_checkpoint_reader import CheckpointReader
SystemError: initialization of _pywrap_checkpoint_reader raised unreported exception
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

[glad zealot]

fck

[brittle wing]

hmm

[glad zealot]

ye no clue

[brittle wing]

ok

[glad zealot]

@sand irisknow it?

[sand iris]

tensorboard freaking out ? idk

[brittle wing]

leave it. i closed

[whole lynx]

guy how to do voice convertion online?

[glad zealot]

-colab

[azure marsh]

Suggestions for @whole lynx

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[whole lynx]

yo thankss

[whole lynx]

finally i can make that one anime girl say my name for the next 10 min while i do the deed

IM KIDDING

[glad zealot]

Ummm

[brittle wing]

100 epochs completed 🎉

[olive crow]

UPDATE:

2 hours of clean mp3 audio, 17 hours as you can see. Goal is to reach 1000 epochs.

[stiff goblet]

Is 9 minutes dataset is too long for OV2 pretrain ?

[whole lynx]

guyyy how do i upload my .pth to huggingface so i can use the online conversion?

[rare goblet]

Ayo? @whole lynx level 3 !!!

[whole lynx]

how do i upload my .pth to huggingface so i can use the online conversion?

[glad zealot]

you need to zip the pth and index

[whole lynx]

whereeee do i upload it? like it soo confusing using huggingface

[olive crow]

do you guys think im overtraining it?

[whole lynx]

is there like a yt tutorial for this?? i need a visual guide like damn huggingface is so confusing to use

[glad zealot]

what was the command to uplaod again

-huggingface

[azure marsh]

huggingface.co upload Guides

English guide

Link: https://rentry.org/fdg_guide_newer
Credits: FDG

Spanish Guide

Link: https://rentry.org/AI_Hub_guide_esp
Credits: Julia (ailen2091)

Hindi guide

Link: https://rentry.org/AI_hub_guide_hi
Credits: Enes

Turkish guide

Link: https://rentry.org/AI_hub_guide_tr
Credits: Enes

[glad zealot]

ah there we go

[whole lynx]

i love you bro

[whole lynx]

wait what a index file? i only have pth

[glad zealot]

it should be in the logs folder

but if you're gonna use it for youself you dont really need the index file

you can just zip the pth and upload it

[whole lynx]

ohh alr thanks

[whole lynx]

that make it way easyer

yea i cant spell easyer lol

[proper shale]

hmmmm idk, check mel again

[olive crow]

ok

@proper shale

[proper shale]

id say keep going

[olive crow]

aight

[stiff goblet]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[tacit grail]

how do I make Applio pronounce words more accurately when it gets to a lower tone? My recent cover is great and all but in the official audio the singer doesnt sing the words distinctively so the last letter in words sometimes doesnt get said after inference.

[whole lynx]

bro i mayyy sound retarted but what the hell do i do next??

like like

i did upload it

now im on the ui

and got no clue on what to do

[glad zealot]

everything you need is in here

[royal trout]

how do i create ai model?

[whole lynx]

o that gonna be a pain bro

i just spented the last 5 hour trying to do it

[royal trout]

me to

[whole lynx]

and now when i wanna make a song with it

MORE ERROR

[brittle wing]

google colab giving 220 epochs in 1 single run. Unbelievable!

[odd shale]

If it gave you 220 epochs, it's because your dataset wasn't that big

[thin stump]

Is 5000 steps normal for a 17 min dataset?

[odd shale]

It depends on the settings you put when training.

Like batch size and epoch counts

[olive crow]

guys do you think its overtrained ?

[brittle wing]

15 files. 15 mins

time to get the model maker role

[odd shale]

I'm not sure

Maybe a bit.

[amber zephyr]

is any one having issues with the feature extraction on rvc disconnected colab?

[solar carbon]

Python path configuration:
PYTHONHOME = (not set)
PYTHONPATH = (not set)
program name = 'env\python.exe'
isolated = 0
environment = 1
user site = 1
import site = 1
sys._base_executable = 'D:\ApplioV3.0.7\env\python.exe'
sys.base_prefix = ''
sys.base_exec_prefix = ''
sys.platlibdir = 'lib'
sys.executable = 'D:\ApplioV3.0.7\env\python.exe'
sys.prefix = ''
sys.exec_prefix = ''
sys.path = [
'D:\ApplioV3.0.7\env\python39.zip',
'.\DLLs',
'.\lib',
'D:\ApplioV3.0.7\env',
]
Fatal Python error: init_fs_encoding: failed to get the Python codec of the filesystem encoding
Python runtime state: core initialized
ModuleNotFoundError: No module named 'encodings'

Current thread 0x00002538 (most recent call first):
<no Python frame>

Hi, may i ask what i need to do with this error in "run-applio" console? Thanks.

[olive crow]

how about now?

[odd shale]

A bit overtrained.

[brittle wing]

Im having an issue running easy gui with rvc local and it keeps giving me an error that says:
Traceback (most recent call last):
File "C:\RVC1006Nvidia\EasierGUI.py", line 22, in <module>
from i18n import I18nAuto
ImportError: cannot import name 'I18nAuto' from 'i18n' (unknown location)
Can someone help me? Thanks.

[olive crow]

should i stop?

[odd shale]

yea

[olive crow]

my goal is to reach 1000 epochs but its at 445

do you know how i stop because the thing im using is doesnt have a stop training button

or i can't find it?

[latent kettle]

why you want to do that?

[olive crow]

because i heard the more epoches you have the better?

isn't it that tho?

[latent kettle]

that is a myth

[olive crow]

oh

so what does epochs mean?

[latent kettle]

yep ....

[latent kettle]

it is a kind of unit to measure a training cycle

[brittle wing]

how can i create audio sample to upload in #1159514067187277865

[olive crow]

whats does it mean by this?

[latent kettle]

for example.. if you read a book for 1 ti.e that means it is 1 epoch

[olive crow]

oh

i see

[amber zephyr]

d

[latent kettle]

using ilaria RVC but don't post process it

[brittle wing]

local/colab?

where can i get voice?

to convert

[latent kettle]

yep .. the quality of your model depends on your dataset quality length as well as on mel, kl, total G and D loss

[olive crow]

are you sure i should stop?

[latent kettle]

do you want to upload your model in #1159514067187277865

[latent kettle]

yes... it's overfitting

[olive crow]

ok

[latent kettle]

yep .. the quality of your model depends on your dataset quality length as well as on mel, kl, total G and D loss

[brittle wing]

Which MDX23C model is better? MDX23C 8K FFT or MDX23C 8K FFT - v2?

It's for the Karafan Colab

[olive crow]

ok i think i just stoped the training

how do i check the models

[latent kettle]

yes .. also read https://rentry.org/RVC_making-models

[latent kettle]

go to assets and then weights

[brittle wing]

Btw i will still using MDX23C 8K FFT.

[olive crow]

yes

and then

[latent kettle]

yes .. now move back and run go-web.bat

[olive crow]

yes and then

[rare goblet]

Ayo? @olive crow level 5 !!!

[olive crow]

@latent kettle

[latent kettle]

click on refresh voice list and index path

[olive crow]

yea

[latent kettle]

select your model from drop down

[olive crow]

and thats it?

[latent kettle]

copy the path of the audio you want to convert

and paste it!

[latent kettle]

here

[brittle wing]

yes

[brittle wing]

how do u know?

[latent kettle]

do you have a model which you have trained?

[olive crow]

but i want my voice model to be so when i talk it talks

real time voice changer

[brittle wing]

yes

[latent kettle]

real-time ?

[olive crow]

yes

[latent kettle]

#🔍│help-w-okada

ask here

[brittle wing]

does the audio sample needs to be in english??

[latent kettle]

https://rentry.org/RVC_making-models
read it

[brittle wing]

no thanks

[brittle wing]

..

[olive crow]

but where do i find the model i trained?

because i dont see it

[brittle wing]

doesnt exist

@sand iris will hindi work in #1159514067187277865 ?

[latent kettle]

https://aihubdocs.github.io/en/extra/model-maker-role/#model-maker-role

[brittle wing]

no info about lang

[sand iris]

Yeah, there are models of various languages

[olive crow]

but where do i find the model i trained?
because i dont see it?

[latent kettle]

are you from India..?

[brittle wing]

yes

[latent kettle]

read it carefully

[latent kettle]

can I dm you ?

[brittle wing]

ok

[brittle wing]

to get the model maker role?

[latent kettle]

you have to know about tensor board

[sand iris]

yeah

[olive crow]

i am in tensor board

[brittle wing]

can i upload a song?

[sand iris]

no

only converted vocals

[brittle wing]

ok

[latent kettle]

without music

[thin stump]

So 5000 steps gave me 120 epochs. That shouldn't be right, going to restart

[brittle wing]

👍

[latent kettle]

or any post processing

[brittle wing]

ok

[latent kettle]

can I dm you?

[latent kettle]

you will see some graphs ...

[rare goblet]

Ayo? @latent kettle level 20 !!!

[brittle wing]

ok

[olive crow]

but how do i download the model because i cant see it in weights

[latent kettle]

which model

[olive crow]

the pth and index?

[latent kettle]

you have trained your model locally?

[olive crow]

idk

yes locally

yes locally i did it locally

[latent kettle]

so why you want to download your model

you have it in your drive

[olive crow]

which drive how can i find it

[teal ridge]

hi, i justo download it and it works on the apk, but how do i make it work in discord or any other apk

like the voice work but idk how to use it on disc

[latent kettle]

here is your model I think

[odd shale]

You can find it on the "weights" folder

[latent kettle]

yeh I told him/her but he/she didn't understand that

[odd shale]

Language barrier

[latent kettle]

look above attachment.. he /she is just there

[brittle wing]

till when will be my model get approved/rejected @sand iris

[sand iris]

When those in charge take a look at it

[brittle wing]

cant u?

[sand iris]

Nope

[brittle wing]

😦

can u give me feedback after listening to it atleast

@sand iris

[sand iris]

heading home

can't listen

[brittle wing]

ok

[flint vapor]

I listened, from the sample look so so but qc will review better

Unfortunately i'm not anymore qc

I used to be in august

[brittle wing]

why u got removed?

[flint vapor]

Idk, after that the first server got deleted i didn't had the role anymore 😢

[brittle wing]

oh okay

[visual pumice]

im begging everyone to set this to 0 (it defaults .25) it makes your cover sound more natural when it comes to volume

[brittle wing]

Yes, it is called "Dynamic Range".

[calm jackal]

-rvc

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[calm jackal]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[calm jackal]

-audio

[azure marsh]

📚 Audio Guides

• Perfecting Audio Isolation on Low-End Rigs: A Practical Guide, by Litsa The Dancer and Faze Masta Google Docs
• Gathering and Isolating Audio, by SCRFilms Google Docs
• Creating Datasets for RVC using iZotope RX, by LUSBERT Rentry
• Vocal Mixing Tutorial, by Roomie YouTube

🛠️ Tools

Audio Separation/Isolation

• https://mvsep.com/
• https://multimedia.easeus.com/vocal-remover/
• MDX-NET Colab

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

[half cove]

@calm jackal if your testing commands please use in #🤖│bots so it doesnt clog chat

[urban grove]

how to install applio on amd 6650 xt
can someone help me

[half cove]

i dont think applio supports dml

[shrewd seal]

INFO:ae:Train Epoch: 1 [0%]
INFO:ae:[0, 0.0001]
INFO:ae:loss_disc=4.861, loss_gen=3.970, loss_fm=11.299,loss_mel=22.474, loss_kl=8.287
DEBUG:matplotlib:matplotlib data path: F:\RVC1006Nvidia\runtime\lib\site-packages\matplotlib\mpl-data
DEBUG:matplotlib:CONFIGDIR=C:\Users\Uc.matplotlib
DEBUG:matplotlib:interactive is False
DEBUG:matplotlib:platform is win32
INFO:torch.nn.parallel.distributed:Reducer buckets have been rebuilt in this iteration.
DEBUG:infer.lib.train.mel_processing:min value is tensor(-1.0833)
DEBUG:infer.lib.train.mel_processing:min value is tensor(-1.0733)
Process Process-1:
Traceback (most recent call last):
File "multiprocessing\process.py", line 315, in _bootstrap
File "multiprocessing\process.py", line 108, in run
File "F:\RVC1006Nvidia\infer\modules\train\train.py", line 271, in run
train_and_evaluate(
File "F:\RVC1006Nvidia\infer\modules\train\train.py", line 473, in train_and_evaluate
wave = commons.slice_segments(
File "F:\RVC1006Nvidia\infer\lib\infer_pack\commons.py", line 51, in slice_segments
ret[i] = x[i, :, idx_str:idx_end]
RuntimeError: The expanded size of the tensor (12800) must match the existing size (0) at non-singleton dimension 1. Target sizes: [1, 12800]. Tensor sizes: [0]

I'm getting this after I split my audio into 10 seconds segment each

but if I don't split it, it gets stuck on epoch 1 forever at INFO:torch.nn.parallel.distributed:Reducer buckets have been rebuilt in this iteration.

[sand iris]

don't split it

I'm assuming the audio segments are too short, and it can't even use them

[shrewd seal]

what about it getting stuck

ohh

[sand iris]

What's your GPU ? And how much time worth of audio are you feeding it ?

[shrewd seal]

gtx 1650 super

[brittle wing]

whats kaggle. is it an alt of colab?

[sand iris]

oh

that would be why

[latent kettle]

AMD

[shrewd seal]

I used audacity to trunicate silience

[shrewd seal]

sso I need to upgrade

[sand iris]

cards with under 6GB of VRAM end up falling back to system RAM

and that's as slow as CPU training which is...

horrendous

[sand iris]

yeah kinda the same thing

[brittle wing]

free?

[sand iris]

I don't remember

[latent kettle]

Why AMD is trending for today 😆

[shrewd seal]

I don't remember how long it took but I have trained another model before

[brittle wing]

https://www.kaggle.com/code/twiced/rvc-v2-kaggle-tested-16-09-2023 lol it was tested on my bday

[shrewd seal]

is amd worth it for ai stuff lol

[latent kettle]

yes. it is something like that

[shrewd seal]

thought it lacked more support

[sand iris]

On windows it's straight ass at the moment

[latent kettle]

no .. I don't think

[sand iris]

On linux it's a mess, but it works at least ? There are issues though

[shrewd seal]

what do u guys think of the 4070

there's one for like 650$ rn

[rare goblet]

Ayo? @shrewd seal level 2 !!!

[latent kettle]

it is pretty good

[latent kettle]

rtx 4070 ?

[shrewd seal]

yes

12gb vram tho

[latent kettle]

I think It is pretty decent gpu more than enough...

[shrewd seal]

I'll think about it

[brittle wing]

is there any guide for kaggle?

oh wait. i have a laptop from 2008 with amd card. will rvc mangio run?

[true badger]

hi! idk if this problem goes here but does anyone know how to fix this?

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
elevenlabs 0.2.27 requires pydantic>=2.0, but you have pydantic 1.10.14 which is incompatible.

GUI never made this error for me before

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[brittle wing]

Absolutely, this highlights a common challenge when working with Python packages and dependencies. Let's break down the error and explore solutions.

Understanding the Error

• Dependency conflict: Your project requires "elevenlabs" version 0.2.27, which needs a newer version of "pydantic" (at least 2.0). However, you currently have an older, incompatible version of "pydantic" (1.10.14) installed.

Resolving the Conflict

Here are a few strategies to address this situation:

1. Upgrade "pydantic": The most direct way is to upgrade your "pydantic" installation:

   pip install --upgrade pydantic
   

   This will usually bring pydantic to a version compatible with "elevenlabs".

2. Pin the "elevenlabs" Version: If upgrading "pydantic" might break compatibility with other parts of your project, you could temporarily downgrade the "elevenlabs" version. Find an earlier version compatible with your current "pydantic":

   pip install elevenlabs==<a-compatible-version> 
   

   You can often find dependency requirements in the project's documentation.

3. Virtual Environments: Create a separate virtual environment for your project. This isolates dependencies, allowing you to use the necessary versions without causing conflicts in your global Python installation. Here's how:

   python -m venv my_env  # Create environment
   source my_env/bin/activate  # Activate it
   pip install elevenlabs pydantic # Install what you need 
   

Important Considerations

• Thorough Testing: After making changes, always test your project carefully. Upgrading packages can sometimes introduce unexpected issues.
• Compatibility: Be mindful of compatibility requirements for other packages in your project when making changes.
• Requirements File: If you have a requirements.txt file, update it after dependency changes to ensure reproducibility of your project environment.

[rare goblet]

Ayo? @brittle wing level 7 !!!

[brittle wing]

@true badger this will certainly help

[half cove]

did you just copy and paste chat gpt?

[odd shale]

Why you did copy a entire chatgpt description?

[true badger]

there was a force update in the colab but now i have two other error

[glad veldt]

what should I set my sample rate as?

[thin stump]

Play it safe with 32k. My model wasn't a true 40k after cleaning it

[proper shale]

use 32k, always

[odd shale]

32k is GOD

[odd shale]

Send screenshot

[dawn nebula]

yeah

[outer oracle]

im traning on 1650 mobile 🔥

[sand iris]

Bruhjak

[olive crow]

does anyone know where the .pth for the model is because i can't seem to find it unlike the .index that was easier to find

[dawn nebula]

assets/weights

[stiff goblet]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[olive crow]

yo but for some reason it doesn't apper

[dawn nebula]

have you finished training the model?

[olive crow]

wait to sec

i might have or not ima dobble check

yep i have

but still doesn't apper

ill send screenshots you can look for yourself

heres index

nothing

im using the https://aihubdocs.github.io/en/rvc/local/mainline/

guide

and idk what the problem is

@dawn nebula

[odd shale]

That issue never happened to me altho i use Paperspace.

[dawn nebula]

scroll down

[olive crow]

ok

there @dawn nebula

@odd shale

[odd shale]

I don't see any .pth file.

[dawn nebula]

its not finished

[olive crow]

its not

[dawn nebula]

continue training it

[olive crow]

then how do i finish it

[dawn nebula]

how many epochs did you train

[olive crow]

450 something

and i set it to 1k

[olive crow]

yo

@dawn nebula

[rare goblet]

Ayo? @olive crow level 6 !!!

[dawn nebula]

rmm

emm

try coninute training it to 1000epochs

[olive crow]

thats gonna take another 18hours

but ill do it

[dawn nebula]

how many epochs have you trained now

450?

yeah set that to 450

[olive crow]

ok so do i run it to 1k

?

because i was told of someone that i should stop the program because i was overtraining

thats why i stopped

[dawn nebula]

you forgot to save checkpoints

[dawn nebula]

no

set epochs to 450

[olive crow]

no i have the check points

how do i do that

[dawn nebula]

no you dont

if you did

you have .pth files in assets/weights

[olive crow]

oh

ok so do i start the program again?

and how do i set it to 450 epochs do i need to restart from scrath?

[dawn nebula]

you dont need to

just put the same experiment name as your previous training stuff

and set epoch to 450

[odd shale]

Maybe he didn't had any .pth files because he didn't mark the "save checkpoint each epochs" option

[dawn nebula]

yeah

[sly pilot]

hey, is this rvc good? "https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases"
if so, how can I train a model?

[odd shale]

Well, that's the original RVC.

There are other versions too

Like Mangio Fork or Applio

But the difference is not that higher

Altho you'll likely need to install tensorboard.

[sly pilot]

oh yeah I remember downloading this before "https://huggingface.co/MangioRVC/Mangio-RVC-Huggingface/tree/main"
is it outdated yet?

[rare goblet]

Ayo? @sly pilot level 2 !!!

[humble blaze]

can someone send em the hina modifed google collab thingy

[odd shale]

It hasn't been update in a long time but you can use it.

[humble blaze]

i cant find it nowhere

[odd shale]

AiCoverGen or W-Okada?

[humble blaze]

w-okada

[odd shale]

Who is the one who wants those colabs? You?

-realtime

[azure marsh]

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

[humble blaze]

-realtime

[azure marsh]

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

[humble blaze]

what the link

[frigid ermine]

  File "G:\Artificial_Intelilgence\Download\RVC1006Nvidia\runtime\lib\site-packages\gradio\routes.py", line 321, in run_predict
    output = await app.blocks.process_api(
  File "G:\Artificial_Intelilgence\Download\RVC1006Nvidia\runtime\lib\site-packages\gradio\blocks.py", line 1007, in process_api
    data = self.postprocess_data(fn_index, result["prediction"], state)
  File "G:\Artificial_Intelilgence\Download\RVC1006Nvidia\runtime\lib\site-packages\gradio\blocks.py", line 953, in postprocess_data
    prediction_value = block.postprocess(prediction_value)
  File "G:\Artificial_Intelilgence\Download\RVC1006Nvidia\runtime\lib\site-packages\gradio\components.py", line 2076, in postprocess
    processing_utils.audio_to_file(sample_rate, data, file.name)
  File "G:\Artificial_Intelilgence\Download\RVC1006Nvidia\runtime\lib\site-packages\gradio\processing_utils.py", line 206, in audio_to_file
    data = convert_to_16_bit_wav(data)
  File "G:\Artificial_Intelilgence\Download\RVC1006Nvidia\runtime\lib\site-packages\gradio\processing_utils.py", line 219, in convert_to_16_bit_wav
    if data.dtype in [np.float64, np.float32, np.float16]:
AttributeError: 'NoneType' object has no attribute 'dtype'```

[sly pilot]

yeah I'm especially looking for a local rvc where you could train

[humble blaze]

nvm

got it

https://colab.research.google.com/github/hinabl/voice-changer-colab/blob/master/Hina_Modified_Realtime_Voice_Changer_on_Colab.ipynb

[frigid ermine]

whats up with this error all of the sudden @sand iris

i had this error before

this mf wasted around 30 hours of my time researching to fix it, even modifying the source code of the projects

but apparently its just cuz the temp file of the adiuo is "None" and theres nothing i can do about it

whys that

[odd shale]

There are a plenty

Just pick between Applio, OG or Mangio

[sand iris]

audio file failed to load

[sly pilot]

I'd choose mangio since that's what I used

[frigid ermine]

yeah but why

[sand iris]

idk

[frigid ermine]

i tried with both mp3 ( the first time ) and then .wav

like i put it through UVR

maybe that messed up smth?

what do i do?

[sand iris]

not sure there

tHONK

[frigid ermine]

im so fucking lucky

it randomly worked this time

[half cove]

rvc moment

[frigid ermine]

i didnt change anything, just retried and it worked ( after 2 times of not working )

so yeah lets hope it doesnt happen again 🥲

( it defenitely will )

[valid hamlet]

how can i hear my self on discord?

ont est

[frigid ermine]

go to audio section

[rare goblet]

Ayo? @frigid ermine level 5 !!!

[frigid ermine]

and click "Lets check"

[valid hamlet]

i do

i mean iam hearing my normal voice

[frigid ermine]

itll play the input to the output

[frigid ermine]

change ur mic then

[frigid ermine]

your mic should be the virtual audio cable

not the normal mic u have

[valid hamlet]

oh

[half cove]

you need to use a virtual audio cable

[valid hamlet]

like this?

[frigid ermine]

yeah like that

👍

[valid hamlet]

getting no voice

[half cove]

what is your output on okada

[frigid ermine]

also change it in output of okada too

[valid hamlet]

this?

[frigid ermine]

the output of okada should be same as input of discord

bro no

[half cove]

no

[frigid ermine]

the output

not the input

[valid hamlet]

oh

[half cove]

input ur mic

output the vac

[frigid ermine]

the input should be ur mic

the output should be the virtual audio cable

[frigid ermine]

yeah

now it should work

[valid hamlet]

it does hmm where can i find a better voice model

[half cove]

#1175430844685484042

[odd shale]

Voice models channel.

It's matter of testing

Or making a model yourself.

[valid hamlet]

how