✨│ai-help

[glad zealot]

yes

[verbal oasis]

and use it as a link?

[glad zealot]

yee

[gray raptor]

what's the best rvc to voice ai songs

[verbal oasis]

can I change the name od them?

[brittle wing]

Help I get an Error in Step 2

[glad zealot]

yup if its just renaming you can do anything with it

[glad zealot]

depends on you, you can check in #1175430844685484042 for voices that you like

[gray raptor]

ok thanks

rip i can't send screenshots

[rare goblet]

Ayo? @gray raptor level 1 !!!

[verbal oasis]

I know only the theory I've never done it before.
How do I load it up on huggingface?
I am in my account

I clicked on + New Model

gave it a name

Licensed it

[glad zealot]

yup

[verbal oasis]

and now I'm here

[glad zealot]

go to files and press the create button

[glad zealot]

yes that one

[verbal oasis]

?

[glad zealot]

yes upload

then just put your zip file

[verbal oasis]

and I upload the zip

noice

There is no save button?

Should I hit Commit changes to main

?

[glad zealot]

yes press comit

[glad zealot]

commit changes

[verbal oasis]

yeah now uplading I guess

Done

Ok

Do you know how can I gt the link?

jeeez I uploaded a .rar

is it the same?

[glad zealot]

not sure if it works with rar

since most RVC stuff uses zip

[verbal oasis]

ok I'll upload a zip then.
After that is this the correct way to get the link?

[glad zealot]

no

right click the small download icon next to the model file

[verbal oasis]

but that will downlad it

[rare goblet]

Ayo? @verbal oasis level 4 !!!

[glad zealot]

right click

then copy link

[verbal oasis]

aaaaah

thankksssssssssssssssssss

time to see if I trained a model correctly

[gray raptor]

@glad zealot after i found the model what else do i have to do

[verbal oasis]

It's working!!!

Thanks for our help

[glad zealot]

wdym

[verbal oasis]

in general how can I get a nice result?
I mean what are the parameter for a high quality audio model?

[glad zealot]

clean dataset, no backround audio or other people talking

[glad zealot]

download it

[verbal oasis]

length of the audio file too?

[gray raptor]

done

[glad zealot]

around 10 mins should be fine

[gray raptor]

then what

[glad zealot]

what are you gonna use it for? voice changer? inference?

cover?

[gray raptor]

cover

[verbal oasis]

and what out these params

are important?

also these

[glad zealot]

use rmvpe_gpu

best reult atm

[gray raptor]

what should i use

[glad zealot]

the lower the batch the closer it will sound like your dataset, so if you're not really that confident about your dataset, you can set it higher to aroung 6-8

[glad zealot]

for cover making?

[gray raptor]

yes

[glad zealot]

are you gonna run it locally or online?

[verbal oasis]

and what about epochs

[gray raptor]

just for myself

[glad zealot]

that depends on you, the more it is the more it train/longer it will train

[glad zealot]

yes but are you gonna run it on your computer or online? of you have a decent gpu, you can run it locally/on your computer, if not run it online on something like google colab

[gray raptor]

hmmm

les try locally

[brave garnet]

RVC Colabs and Spaces

⠀

Local Forks 🖥️

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

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

RVC Studio
Simplified, suggested for all, by SayanoAI.

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

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

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

Made with love by Ilaria 💖

[verbal oasis]

so 150 is a good number?

[glad zealot]

id personally go 200 but 150 is fine

[gray raptor]

sooo what should i do @glad zealot

[glad zealot]

select one of those for your liking

im going to bed now :D

[gray raptor]

huh

[brave garnet]

RVC Colabs and Spaces

⠀

HuggingFace Spaces 🤗

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

RVC-HFv2
Applio port, by r3gm.

AICoverGen
AICoverGen port, by r3gm.

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

Made with love by Ilaria 💖

Voicechanger Download (Okada)

⠀
Download for Nvidia GPUs
Version 17b cuda

Download for AMD GPUs
Version 17b directml

Download for Intel GPUs
Version 17b directml

Download for Mac
Version 17b Mac
⠀

Made with love by Ilaria 💖

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs

F0 Det.: rmvpe (suggested for all series)

RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra

Advanced Settings

Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low

⠀

Made with love by Ilaria 💖

RVC Colabs and Spaces

⠀

Google Colabs

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

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

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

EasyGUI
The OG interface, by Rejects.
⠀

Made with love by Ilaria 💖

[rare umbra]

Does anyone know the reason why it sounds like it's demonic? 🗿

I use this and... I don't know how to make the original voice not sound, or make it sound like there are two voices

[viscid moss]

Outdated colab, try the new ones

-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

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

[viscid moss]

U can find guides here:
https://docs.aihub.wtf/

[chilly spade]

are here some german models?

[rare goblet]

Ayo? @chilly spade level 1 !!!

[chilly spade]

?

[naive violet]

what's the difference between rvc and rvc v2? also, how do i use a rvc v2 model?

[rare goblet]

Ayo? @naive violet level 4 !!!

[proper barn]

How do I send a screenshot here?

[sand iris]

you should be able to now

[sand iris]

models were altered slightly, that's about it

unless you're using RVC-GUI, the software you're using must support both

[naive violet]

alright!

[proper barn]

ok lemme send

whenever i try to load in a model this is the message i get

[low shard]

You are using an outdated colab

Use only the uoddated ones in our docs

https://docs.aihub.wtf

[visual tiger]

Yo bros, is the RVC V2 model still working?

[proper barn]

is there an updated version of colab?

[rare goblet]

Ayo? @proper barn level 1 !!!

[low shard]

In the docs theres an uupdated version of easygui, so yes and there are also other colabs too

Check it out

[proper barn]

ok tysm!!

[low shard]

Yww

[low shard]

Yes rvc still works, do u got any errors ?

[visual tiger]

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
This one

[whole ruin]

hiiii i had a question!
for the voice changer
does it use python by chance?
cuz i delete the python i had
and moved to 3.10.11
and now i can see that the voice changer is not being accurate or just not picking up well

[visual tiger]

I just want to make sure because my friend said it wasn't working for her

[verbal oasis]

is there a fix for this?

Unfortunately, there is no compatible GPU available to support your training.

[low shard]

Yea that colab is updated and works

Whats the error

[low shard]

For the wokada voice changer please ask in #🔍│help-w-okada or #1192011222023950368 as rvc is the tool to make ai covers

[visual tiger]

Thanks

[rare goblet]

Ayo? @visual tiger level 1 !!!

[visual tiger]

No no, there is no error, I'm just asking based on what my friend said

In fact I'm still new to it

So I'm still trying to figure out how to use it

:|

[low shard]

Check our docs for a deeper tutorial of it

https://docs.aihub.wtf

[verbal oasis]

anyone know how to force allocate a GPU to train a model?

[visual tiger]

Gotcha, thanks for the assist

[low shard]

Yw

[hard crystal]

speaking of which, i was taking a look at the readme and saw this mentioning RVCv3, is there any updates or progress on it? because i heard it was being developed quite a while ago (~6months??)

High quality licensed song datasets will be added to the training-set often for your use, without having to worry about copyright infringement.

Please look forward to the pretrained base model of RVCv3, which has larger parameters, more training data, better results, unchanged inference speed, and requires less training data for training.````

[whole ruin]

damn no one is like ever active there

kinda sad ngl

[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

[tired escarp]

how to make a voice model

[iron cobalt]

RVC Disconnected unexpectedly disconnects from Google Colab when trying to train. Does anyone else have this problem?

[tired escarp]

me

[west agate]

Would anyone know how to remove gameplay from a audio? i have tested many different ways and can't seem to figure it out. Any help would be appreciated :)

[woeful nexus]

windows won't let me run mangio 😦 what can I do?

ok nvm I did it

[low shard]

Check https://docs.aihub.wtf

[low shard]

Check #1213586188091523075

[tired escarp]

error 404

[rare goblet]

Ayo? @tired escarp level 1 !!!

[low shard]

When i click the link i get no error 404, retry

[tired escarp]

when click "How to Make Voice Models"

[low shard]

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

Retry

[tired escarp]

thx

[west agate]

@low shard would you know anything about this?

[silent cloak]

Does anyone know how i can make custom voice models via Google Colab? All the guides i found on YouTube are mostly outdated and not working, if anyone does i want a step-by-step guide about this, the AI Hub guide didnt really help much

[low shard]

there is a guide to clean up audio in https://docs.aihub.wtf other than the ones in the guide nope

[low shard]

check the rvcdisconnected guide in https://docs.aihub.wtf/

[rare goblet]

Ayo? @silent cloak level 1 !!!

[snow dome]

I'm having some trouble with AICoverGen: AttributeError: module 'pkgutil' has no attribute 'ImpImporter'. Did you mean: 'zipimporter'?
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error

[rare goblet]

Ayo? @snow dome level 1 !!!

[snow dome]

It happened when I'm trying to install: pip install -r requirements.txt

[strange quest]

-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

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

[lyric vine]

How turn nvidia off

[low shard]

dont, you need the free daily gpu to train ai models, click the arrow next to ram and choose the t4 free gpu

you have around 4 hours for free daily

[half cove]

they aint training

but yea use t4

[low shard]

vertical images look so small on pc 😭

i feel the pixels

[half cove]

click on the image

🗣️

[low shard]

i did 😭

[lyric vine]

No work

[rare goblet]

Ayo? @lyric vine level 3 !!!

[lyric vine]

Man i am unlucky boy ever

[hollow sluice]

who help me to install on mac(i am french and i dont speak english)

[brave garnet]

RVC Colabs and Spaces

⠀

Local Forks 🖥️

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

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

RVC Studio
Simplified, suggested for all, by SayanoAI.

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

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

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

Made with love by Ilaria 💖

[sand iris]

J'aimerais bien, mais j'ai pas de mac sous la main pour test

et c'est globalement un enfer vu que je sais pas comment ça se comporte

[hollow sluice]

okada et rvc c le meme logiciel??

[rare goblet]

Ayo? @hollow sluice level 1 !!!

[hollow sluice]

ta pas le lien du telechargement?

[sand iris]

Non

Tu veux quoi ? RVC pour convertir des fichiers ou okada pour du temps réel ?

[hollow sluice]

bah un voice mode en temp reel pour mac

et gratuit

[sand iris]

C'est lequel ton mac ?

[hollow sluice]

macbook pro m1 2020

[sand iris]

Oh nickel ça

https://huggingface.co/wok000/vcclient000/blob/main/MMVCServerSIO_mac_onnxcpu-nocuda_v.1.5.3.17b.zip

[hollow sluice]

et sa sa bug pas

[sand iris]

Y'a de la latence

[hollow sluice]

ah mais je l'ai deja

[azure marsh]

Suggestion for @mellow flax

🍏 Applio-RVC-Fork

Enhanced RVC Variant: Optimized Performance Through Modifications, Built upon Mangio-RVC-Fork.

[hollow sluice]

et sa fait des audio chelou avec la voix de macron par ex

[low shard]

did you put the t4 gpu?

[hollow sluice]

en plus sa me fais sa et quand je fais reload sa remet

[sand iris]

y'a quoi dans le terminal ?

y'a des erreurs habituellement

[barren quartz]

am i able to fix... this?

[lyric vine]

Yes but he say 'no work can conncet without gpu?'

[low shard]

click the same arrow, click disconnect and delete runtime and re run all cells

[stray mica]

What's the point of thisc hannel if I can't send an iamge

[rare goblet]

Ayo? @stray mica level 1 !!!

[stray mica]

oh now I can

why does nothing show

[low shard]

what google colab u using

[stray mica]

I refreshed it and it worked

[eternal dawn]

Dont know if this colab is outdated or not, but when I try to download the model an error pops up. This is the colab: https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=7vh6vphDwO0b

[lyric vine]

Keep same error

[low shard]

it is, check https://docs.aihub.wtf

[snow dome]

Is there any standalone app for RVC or something?

[low shard]

sorry but idk much of aicovergen, maybe @viscid moss could know why u still get it

[eternal dawn]

Thanks for telling, any google colab I can use (suggestion)?

[low shard]

i personally use ilaria rvc for making ai covers
rvcdisconnected for making ai models
applio is good too and can do both with a user friendly interface so its your choice tbh

[eternal dawn]

Thank you brother/sister.

[low shard]

yw

[spare shale]

This cover turned out well for me

Well I mean that sounds good

[thin stump]

yeah either you extract the backing vocals out and convert it through rvc, or leave it alone and add it back in

Try BVE

[viscid moss]

Can u send a screenshot?

[viscid moss]

https://tenor.com/view/usseewa-gif-23531543

[eternal dawn]

I have a question on Illaria RVC. Im kind of confused. When I try downloading the model it says its an error and to check the link if it's valid, but it can also be the download model name I put wrong. This is the huggingface link im trying to use: https://huggingface.co/LilYoda/NarraterHunterxHunter/resolve/main/NarratorHunterxHunter.zip
My questions is how do I use Illaria's RVC properly? I'm so used to the other google colab cause it just automatically placed in the model. Apologize for the interruption

[low shard]

be sure the name has no spaces

[eternal dawn]

Yeah, I did that

[low shard]

check you havent closed the google colab and that you are connected to the t4 free daily gpu

[eternal dawn]

I think I closed it, thanks for telling

I will try Applio too as stated

[low shard]

yea no dont ever close the google colab until you are done

[eternal dawn]

Yeah I used both and both were fine

[barren quartz]

the high note is pretty weird to me tbh Idk how to fix that

[barren quartz]

thank you 🙏

[spare shale]

Check if it is sung well in the original

[thin stump]

yw, if it doesn't work then HP 5 or HP 6 models can catch onto it

[thorn cypress]

Hey uh, I'm trying to make a model of my own voice. But I want to use it for like songs, so would I need to sing in the recording for my model, or would talking just work? Or should I just try both and see which one sounds better?

[hearty bobcat]

how to make a model

help

someone send me the colab link

/colab

[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

[hearty bobcat]

no bitch i cant find the stupid colab link

[rare goblet]

Ayo? @hearty bobcat level 1 !!!

[hearty bobcat]

the goo0d one

[half cove]

-colab

[azure marsh]

Suggestions for @hearty bobcat

☁️ 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

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

[proper barn]

i keep getting this message when i try to run gradio

[lyric vine]

Here but is portugesse bc i am brazil

[viscid moss]

Change to another google acc

[worn pebble]

Hey guys I am new here, i wonder if u guys know if there is trained model based on Emilia Clarke

[simple obsidian]

does anyone know how to fix this? (i have google colabaratory can't check step 2: download the model ai and writes some error file)

[proper barn]

it looks like the one i got.. someone told me it was bcz im using an outdated version of colab

[undone narwhal]

RVC V2 disconected. colab

Preprocessing and Feature Extraction

error

---> 16 gpuList = gpus.split("-")
17 cmd = f"python extract_f0_print.py "{exp_dir}" {cpu_threads} {pitch_extraction_algorithm} {crepe_hop_length}"
18 print(cmd)

NameError: name 'gpus' is not defined
Does anyone know how to improve this error or report this error?
The colab URL is this https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS#scrollTo=Yj-_npuA_HRB

[grim halo]

"Go to Experiments. In Filter type "pasting"

can someone explain where to find experiments

[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

[dusty bramble]

whats the diff between 32 and 40k sample rate?

[thin stump]

Most datasets are 32k anyways https://docs.aihub.wtf/rvc/resources/datasets/#section-9

[spiral osprey]

how to create space in huggingface actually?

[dusty bramble]

if i got my own dataset which should i choose?

[thin stump]

check your spectogram with audacity. Here's an example

It says 20k so I can use 40k as a model, but I still resampled it down to 32k because it'll sound better with less robotic artifacts

[dusty bramble]

im asking because i trained a voice with 32k, then i got more data, but i forgot to put at 32k, and just left it at 40k

and the one with less data sounded better

:v

[thin stump]

Hoping I get this term correct so no one yaps at me 😆 . That's called overfitting (for sampling rate) since rvc tries to guess what's in the missing frequencies with 40k

[dusty bramble]

so, should i just retrain it at 32k again?😭

[thin stump]

for sure

[dusty bramble]

okok, ty

o7

[thin stump]

could be the new audio added too, you never know. Just putting it out there as a possibility

[dusty bramble]

thats what im thinking, i honestly dont know lol, i would train the older version to the quality of the new one to check it out but it takes too long

[main plover]

What is the difference between .pth and .index files?
Because I trained a model in google colab and downloaded the .pth file but it disconnected before I could download the .index file 😦
Should I train again? or I can use any index file?

[rugged stirrup]

Could someone give me the link to download tensorboard venv? cause I cant find it

[lament crest]

is all voice models are rvc v2? i was making ai voice for google colab and i cant make ai voice

[little owl]

.pth files are the model itself while .index files help with the models “accent”. you don’t need to retrain, just go back and load your dataset and train the index and it should be good

[lament crest]

hello guys how can I solve this problem?

[abstract kestrel]

Is there anyone here who can write lyrics and vocals?

[sand iris]

you mean that one script ?

look up "tensorboard script" in the messages on this server, it's somewhere

[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

[unique blade]

why tf do they sound defrrent

thers subbosed to sond like this

[halcyon oxide]

Can anyone help me Ilaria rvc has a 'torch' problem

[rare goblet]

Ayo? @halcyon oxide level 3 !!!

[low shard]

just click runtime, disconnect runtime and delete, and re run everything again

[halcyon oxide]

thank you

[low shard]

Yw

[white hound]

Всем привет
Помогите с установкой люди доюрые

[unique blade]

wtf is this

[lilac gulch]

Hey Guys,

I'm having some trouble with RVC.

Trying to use Camila Cabello RVCv2 [500 Epochs] (https://huggingface.co/AIVER-SE/CamilaCabelloV2/resolve/main/CamilaCabellov2.zip) but I'm getting an error on Step 2. Download The Model

`---------------------------------------------------------------------------
JSONDecodeError Traceback (most recent call last)
<ipython-input-5-2e2a28a5fec2> 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)`

Any idea on what the issue is and how I can find a resolution?

[low shard]

You are using an outdated colab

Check https://docs.aihub.wtf

I personally use ilaria rvc

[lilac gulch]

Thank you boss, much appreciated!

[low shard]

Yw

[proper barn]

I still get this message when I try run gradio

[gusty socket]

Hey everyone! So I downloaded the app and downloaded a model, how do I make them work?

The default models work (with a lot of echoing)

[lilac gulch]

Yep, I'm getting the same error too and I'm watching the tutorial as well to make sure I'm doing it right.
Traceback (most recent call last): File "/content/Ilaria-RVC/app.py", line 1, in <module> import subprocess, torch, os, traceback, sys, warnings, shutil, numpy as np ModuleNotFoundError: No module named 'torch'

@low shard Any idea whats happening here?

[rare goblet]

Ayo? @lilac gulch level 1 !!!

[sand iris]

@spiral chasm How did you set up RVC ?

[static gyro]

hi!

im lerking and ill try the 32k thing

[proven hill]

runtime, disconnect and delete data, refresh the page

[spiral chasm]

i downloaded git repo on ubuntu, installed python3.10 locally and according to read.me from github installed required sh ./run.sh dependencies.

[sand iris]

You haven't touched anything besides that right ?

By default it will be using the CUDA (nvidia) version

[spiral chasm]

yes and this is incompatible

[sand iris]

Also, what is your exact GPU model ?

[spiral chasm]

I have to check, I'll write it down soon

[sand iris]

👍

[tiny totem]

-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

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

[spiral chasm]

is this information sufficient?

$ sudo lshw -C display
description: VGA compatible controller
product: Renoir
vendor: Advanced Micro Devices, Inc. [AMD/ATI]
physical id: 0
bus info: pci@0000:04:00.0
logical name: /dev/fb0
version: c2
width: 64 bits
clock: 33MHz
capabilities: pm pciexpress msi msix vga_controller bus_master cap_list fb
configuration: depth=32 driver=amdgpu latency=0 mode=1920x1080 resolution=1920,1080 visual=truecolor xres=1920 yres=1080
resources: iomemory:fc0-fbf iomemory:fc0-fbf irq:46 memory:fce0000000-fcefffffff memory:fcf0000000-fcf01fffff ioport:1000(size=256) memory:d0500000-d057ffff

dev@dev-Inspiron-5505:~$ lspci
00:00.0 Host bridge: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne Root Complex
00:00.2 IOMMU: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne IOMMU
00:01.0 Host bridge: Advanced Micro Devices, Inc. [AMD] Renoir PCIe Dummy Host Bridge
00:01.2 PCI bridge: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne PCIe GPP Bridge
00:02.0 Host bridge: Advanced Micro Devices, Inc. [AMD] Renoir PCIe Dummy Host Bridge
00:02.2 PCI bridge: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne PCIe GPP Bridge
00:02.4 PCI bridge: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne PCIe GPP Bridge
00:08.0 Host bridge: Advanced Micro Devices, Inc. [AMD] Renoir PCIe Dummy Host Bridge
00:08.1 PCI bridge: Advanced Micro Devices, Inc. [AMD] Renoir Internal PCIe GPP Bridge to Bus
00:14.0 SMBus: Advanced Micro Devices, Inc. [AMD] FCH SMBus Controller (rev 51)
00:14.3 ISA bridge: Advanced Micro Devices, Inc. [AMD] FCH LPC Bridge (rev 51)
00:18.0 Host bridge: Advanced Micro Devices, Inc. [AMD] Renoir Device 24: Function 0

01:00.0 Non-Volatile memory controller: SK hynix BC511
02:00.0 Network controller: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter (rev 32)
03:00.0 Non-Volatile memory controller: Sandisk Corp WD Blue SN550 NVMe SSD (rev 01)
04:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Renoir (rev c2)
04:00.2 Encryption controller: Advanced Micro Devices, Inc. [AMD] Family 17h (Models 10h-1fh)
04:00.5 Multimedia controller: Advanced Micro Devices, Inc. [AMD] Raven/Raven2/FireFlight/Renoir

[sand iris]

This seems to be an integrated GPU

[spiral chasm]

yes

[sand iris]

Training is not possible on those sadly

Unless you are able to allocate 8GB of memory to it (which is really rare)

[spiral chasm]

I have 32GB, so I guess I can - and what minimum parameters should be enough for me? What equipment?
Maybe I'll get...

[sand iris]

The GPU needs to have at least 8GB

for some devices, you can set the integrated graphics memory size

[proper barn]

Is the delete data supposed to say “disconnect and delete runtime”

[misty tapir]

hi can i ask about how to merge pth file?

so like make these pth files to single pth file

[old oracle]

-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

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

[tiny totem]

I feel like this should be pinned so everybody can see it easily

[covert sapphire]

I'm not familiar with coding or anything, but I tried to run the voice changer and it doesn't seem to be working. It says this

Connection aborted Connectionreseterror

An existing Connection was forcibly closed by the remote host

anyone know what I should do?? cuz I'm completely in the dark here 😭😭😭

[rare goblet]

Ayo? @wicked cradle level 1 !!!

[wicked cradle]

plz

i want to download it

[proud elbow]

nah you have to pick which one is best according to the testing, also it tends to be around 200-500 epochs (excuse my subjective thought but still test by yourself)

[wicked cradle]

can i get the link for the voice changer plz

[misty tapir]

thanks, that is perfectly helping

thanks for the advices

[half cove]

ill pin the website instead ig

cuz link changed

** Please read this tutorial before asking any questions**

https://docs.aihub.wtf/

[tight wind]

My TB is looking like this. Am I doing something wrong or should I wait it out till its done

[rare goblet]

Ayo? @tight wind level 1 !!!

[brittle wing]

Hi, when i try to install new model to EasyGUI i always get this popup "Warning. Failed to download file" when I first downloaded it worked but now didn't

[glossy dome]

good evening, sorry, I tried to follow the guide but the one in Italian doesn't work so I followed the one in English but I didn't understand well, perhaps because of the translation, by any chance can someone send me exactly the point where it says how to create a voice AI model from 0 in pth format? using my mp3 voice as an example and uploading it as a template

[brisk salmon]

dang collab have not gpu free, and hugging face is so laggy for ai cover

[low shard]

gpus arent free, and hugging face isn't "laggy" its just slower cus instead of gpu it uses cpu, which is wayy slowerr

[brisk salmon]

oh okay i see

i have an amd gpu does i can run locally ? rx6800xt

[sand iris]

Not on windows for the time being

[brisk salmon]

a sad

[low shard]

check https://docs.aihub.wtf

[lethal shale]

HELPPP

NOTHING HEREE

[shrewd nest]

guys im running applio, i loaded model succesfully
but i wanted to use TTS
i loaded my model files., but it ask me to choose TTS voice (already there are list of inbuilt TTS voice) (why is that, i wanted to use my model voice for TTS) not other inbuilt voices

[torn bluff]

I have decided to use Mangio RVC to train my screaming/growl/singing model. I am watching a YouTube video step by step. For starters it is taking WAY longer than I thought. Hopefully it is still running... it seems to be no changes so far after 10+ minutes (Feature Extraction). I was told I should use the best model available which is called "rmvpe". Im thinking it's because my GPU is old (Geforce GTX 1650). I need to know if it will still finish. I don't mind if it is a 1 hour or 2 as long as I know it's still working and nothing wants to tell me it failed. Also when you train a model from a WAV file can it be multiple styles of vocals in one track or am I supposed to seperate them into multiple WAV tracks? Because as I said at the beginning of this paragraph, im training the same vocalist but he screams/growls and sings in his songs.

[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

[lethal shale]

if you don't send a screenshot we won't be able to help

[torn bluff]

@lethal shale Sure ill send one

@lethal shale

[proud elbow]

either:

• generate built-in TTS then convert using the RVC model
• train your own TTS voice model in GPT sovits

[lethal shale]

The one above it worked and it didn't?

[torn bluff]

The screenshot on the right is my current process. it has been at least 20 minutes. and it says the same thing. I was told it could take a long time depending on my GPU and because i chose the best quality model "rmvpe" that it could also take a while

[lethal shale]

What does the cmd say it is the same?

[torn bluff]

im sorry. what do you mean?

Yeah its the same

[lethal shale]

I mean before extraction there is another process which is just above it

[torn bluff]

ill show you what it would look like if it was done from the YouTube video im watching

Oh you mean this

Yeah its done

[lethal shale]

Normally it should take from 1 minute to 5

[torn bluff]

Even with the best training model for feature extraction?

my GPU is old. could it be that?

[lethal shale]

I'm training a 80mb voice file right now, and it took 3 mins for extraction

[torn bluff]

This is what it would look like if Feature Extraction was done

[lethal shale]

Yes exactly it should look like that

can you check your cpu usage?

[torn bluff]

Sure

[lethal shale]

Is phyton actually processing

we can see it there

[torn bluff]

its running between 60-80%. jumping all over

[lethal shale]

Then it is processing the data but it is strange that you don't get the result

[torn bluff]

Like i said old GPU?

I am running a Geforce GTX 1650

[lethal shale]

I guess you have 4Vram right?

4gb

[torn bluff]

No I have 16 GBs

[lethal shale]

Not ram

Vram

[torn bluff]

oh vram

[lethal shale]

I think 1650 has 4

If it is 4 it should be slow but it should still work

[torn bluff]

It says 8 in system information

[lethal shale]

can you run dxdiag and check?

[torn bluff]

okay sure

[lethal shale]

Even 2080 has 8gb vram

no way 1650 has 8gb

click the third section

it should show the Vram

[proud elbow]

GPU-Z can also show detailed gpu information

[torn bluff]

@lethal shale Found it you are right

[rare goblet]

Ayo? @torn bluff level 10 !!!

[lethal shale]

4gb Vram is slow but not that slow mate

[storm holly]

anyone knows during all the empty spaces, the output has some sorta white noise?

[rare goblet]

Ayo? @storm holly level 7 !!!

[torn bluff]

LOL if I just had more patience

[lethal shale]

Now biggest question

How long is your voice data

[torn bluff]

5:45

[lethal shale]

Good lol

Otherwise you'd wait a day like me

[torn bluff]

i have a good question too

[proud elbow]

batch size?*

[lethal shale]

he should use 4 I guess

4 batch size

[proud elbow]

4 gb vram could just go 2, 4 possible but may get performance bottleneck

[lethal shale]

And I think 100-120 epooch

[torn bluff]

my voice model has screams/growls and singing in one wav file. I got samples from 3 songs. will that screw up the training having multiple vocals styles even if it comes from the same vocalist? Also am I supposed to separate the vocal styles in like 2 or 3 different WAV files?

[proud elbow]

still 4 is better for quality

[lethal shale]

2 would take forever

[lethal shale]

Well that depends

[proud elbow]

it'd also fill out the vram anyway

[torn bluff]

This is the next step you guys are talking about. What do i do next?

[lethal shale]

Try 110 epooch and start with 2 batch size

also

are you using firefox

damn

[torn bluff]

Yes for Mangio and currently Discord

[lethal shale]

We can't see the save frequency

on Firefox

[torn bluff]

For most things I use Brave

[lethal shale]

it is 5 on defaulth

[torn bluff]

It's not telling me even if I hover over the save frequency

[lethal shale]

Well set epooch 110

and start with 2 batch

let's see how it goes

if you get the first epooch in 2-3 mins it should work

The process will take around 6-7 hours

[torn bluff]

What about this part?

[lethal shale]

Leave it like this

[torn bluff]

Okay here we go

[lethal shale]

it should take around 2 to 5 mins for first epooch

[torn bluff]

Sorry actually before I continue... I should have screenshotted this too. I thought it was another step below. My bad

[lethal shale]

that looks fine right now

[torn bluff]

Okay so change nothing?

[lethal shale]

Nope consider it as a test for now

we will see how it is with 2 batch size

[torn bluff]

Model is being trained now

This is going pretty fast. It's already almost 25% done

[lethal shale]

That is just the firt epooch

there are 109 more

[torn bluff]

Ohhh lol

[lethal shale]

lol

this is mine right now

[torn bluff]

Mangio too?

[lethal shale]

Yeah mangio

Did you also make it V2

[torn bluff]

Yup

[lethal shale]

Did you check if it is 32k or 40k

[torn bluff]

40

40 was the lowest option

[lethal shale]

On V2 32k is the lowest option

are you sure

[torn bluff]

oh wait no

32K is the last choice so yeah. Anyways i chose 40

[stone coral]

Help I'm new how do I make ai bots say something?

[torn bluff]

It confused me because you would figure 32K being the smallest sample rate would be the first choice

[lethal shale]

wait we can check if you did it right

https://academo.org/demos/spectrum-analyzer/

Click here and upload your voice file

And click play

If it is around 18k 20k you should set it to 40k

if it is lower you should choose 32k

[torn bluff]

Oh the wav file I chose for it you mean! I thought you meant my sample rate for the AI on Mangio

Sure ill do that

Oof @lethal shale Well that sucks because I had already chose 40K before I trained the model. Isn't it too late now?

It's just the vocals. I don't know if it matters

[proud elbow]

still go 32k

[lethal shale]

Well 32k would fit better here yes but trying won't hurt

How is your cmd right now

[torn bluff]

Okay. I switched to 32K

[lethal shale]

first epooch came?

you need to redo whole process to switch

[torn bluff]

How would you know if the first Epoch came?

[lethal shale]

Send a screenshot of cmd

[torn bluff]

OHHH

I didnt realize my cmd was open

[lethal shale]

if it is closed you can't train lol

[torn bluff]

it's not closed

[lethal shale]

Yeah it is working quite well

[rare goblet]

Ayo? @lethal shale level 7 !!!

[lethal shale]

So I suggest you to go back and restart with 32k do everything else the same

[torn bluff]

This will be a while. When I did it in Google Colabs "EasyGui" it would be at 50 Epochs by now.

So just cancel this and start over?

[lethal shale]

Well if you want you can test it

Maybe it will work well

And if you don't like you can redo again

[torn bluff]

now i don't know what to do lol

[lethal shale]

I think you should let it finish and then you can do with 32k and decide which is best

[torn bluff]

Yeah i think i made the decision to wait and find out

[lethal shale]

Also download Tensorflow

[torn bluff]

?

[lethal shale]

It didn't work for me unfortunately

But it is necessary to know when to stop training

https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/

Overtraining damages the quality and takes your time

[torn bluff]

Okay so what will Tensorflow do?

[lethal shale]

There is supposed to be a graph that tells you when to stop

The guide explains it

[proud elbow]

pip install tensorflow

[lethal shale]

It doesn't work for me

[proud elbow]

it will give better features on the tensorboard

[torn bluff]

i think ill skip it for now. This is just another thing I don't feel like learning how to use right now lol

[lethal shale]

It is stable for me why?

[proud elbow]

looks like nan value issue in mangio with GTX 16-series

[torn bluff]

1650 is part of that series?

[proud elbow]

do u have that issue?

[torn bluff]

I wouldn't know. I don't use Tensorflow or whatever it's called. We are talking about that right?

[lethal shale]

I have RTX

[proud elbow]

see the console log if the loss values are nan

[lethal shale]

is there a way to fix that shit

[torn bluff]

🤷‍♂️

[proud elbow]

still the same Turing architecture, though I noticed it on a 1660 super

[lethal shale]

Yeah lol it is so useless for me and it will take my time so much

I will try all 110 models to see which is better aah

[proud elbow]

and probably not the latest driver now

[lethal shale]

Oh I didn't have my drivers updated

[torn bluff]

Ill just mess around until i get the model I want. I just got to learn patience. Now I know I have unlimited attempts now I moved away from Google Collabs lol

[proud elbow]

I have no clue

[lethal shale]

At least you are not trying to learn Stable Diffusion

You find out 5 things to learn when you learn one

I mean

ComfyUI

[torn bluff]

All I know is the Google Collabs suggestion was pissing me off lol I would get timed out even after almost finishing a model

[lethal shale]

Those schemes haunt me at night

[lethal shale]

Yeah I feel like when done with collab the quality is not that good

[sand iris]

Ehh, it ain't the same

[torn bluff]

Mine was worse then quality. The voice sample it would come back with was just a normal talking voice instead of the screams etc i was looking for

[sand iris]

They gimped the 16xx cards

[torn bluff]

Oh boy, this is going to take a minute... 💀

[lethal shale]

I never tried any screams lol but I hope that one will do

[proud elbow]

yea the nan issue I meant 💀

[lethal shale]

If your sample was 30 mins

that would take 5-6 mins for each I guess

[torn bluff]

it would be hours if it was 30 minutes i would assume 💀

well for me and my GPU

[lethal shale]

At least you will be sleeping but your poor gpu will be the one dealing with the heavy burden lol

The worst thing is that you wait for 12 hours sometimes and it ends up being a massive shithousery lol

[proud elbow]

I yet to confirm if the nan issue doesnt exist in any non-Turing cards at all besides my hallucinating speculation 💀

[torn bluff]

Well my GPU is like if Goku stayed in Super Saiyan Blue for more than 1 hour. If I decided to go to Kyoken times 10 (watch youtube videos in 1080p) or more he will lose his power level

[sand iris]

They're not quite the same arch

And it has only ever happened randomly on 16xx cards

What it could be is a messed up audio file here

[lethal shale]

I have no idea who Goku and where SSB is lol

But I feel you

[torn bluff]

Dragon Ball Super reference lol

[lethal shale]

Ah shit my friend was crying

Someone died there

right

recently

[torn bluff]

what?

Oh you mean the author of the manga?

Yeah he passed away on March 1st

anyways off topic lol

[lethal shale]

I would say

No that would be offensive to him lol

He was the one the march first

[torn bluff]

Im going to watch my YouTube videos in 480 p just to give my GPU a bit of a break

[slate gate]

hey! is Ilaria working? Tried to start it now after installing and clicking "start" it doesn't work no more

[brisk oak]

man i just wanan find a good girl voice

[turbid pagoda]

Yeah I just got a Collab limit warning on llaria that i never had before did google get to that one too?

[slate gate]

hm sounds sad

it shows this mistake with me

[brisk oak]

best free anime/cartoon girl rvc model voice?

[rare goblet]

Ayo? @brisk oak level 1 !!!

[turbid pagoda]

I can go through the entire process but it wont bring up the gradio/Tenserboard

[slate gate]

Do you know any other collabs like Ilaria?

[turbid pagoda]

the only one that worked for me this morning was applio. but you cant use hugging face models

[rugged stirrup]

Ok

[turbid pagoda]

this is really frustrating everything is constantly getting put behind a pay wall. I already pay for kits ai but they dont always have the model i need. plus with kits you have to google the model and hope somebody made it months ago

[proud elbow]

just download them

[torn bluff]

That's why I am making my own model now. The one I am training doesn't exist unfortunately. I have tried different collabs and have failed in multiple attempts. Hopefully Mangio is successful in the model I want

Started at 2:30 and im only at 23 epochs out of 110... It will be like 9 pm EST by the time it's done... 💀

[low shard]

just click runtime, disconnect and delete runtime, and re run it all again

[calm tangle]

so i updated automatic1111's stable diffusion locally and it updated the components that Applio 3.0.8 uses,
now im getting this:

torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 2.00 MiB. GPU 0 has a total capacty of 6.00 GiB of which 0 bytes is free. Of the allocated memory 3.65 GiB is allocated by PyTorch, and 7.87 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

what can i do from here?

[lethal shale]

Does leaving cmd open for 15 hours etc damage VRAM? I run it with low Batch so I only use 65% of gpu

I found no info about it anywhere

[pearl wind]

what to do if it says that use torch contentvec D:\a\_work\1\s\onnxruntime\python\onnxruntime_pybind_state.cc:574 onnxruntime::python::CreateExecutionProviderInstance CUDA_PATH is set but CUDA wasn't able to be loaded. Please install the correct version of CUDA and cuDNN as mentioned in the GPU requirements page (https://onnxruntime.ai/docs/reference/execution-providers/CUDA-ExecutionProvider.html#requirements), make sure they're in the PATH, and that your GPU is supported.

[torn bluff]

I hope not cause my model is only at 30 epochs out of 110 and its been running since 2:30 pm est 💀

[lethal shale]

Mine is on for 8 hours already 💀

I checked the tempatures though nothing wrong

[torn bluff]

That's even worse if it does damange vram 💀

[lethal shale]

I use low batch size to prevent it

Right now it uses only 60%

[torn bluff]

okay good. you told me to use 2 so i should be fine too

[lethal shale]

and temperature is 65 so it looks quite good

I get one epooch in 4 minutes

lol

[torn bluff]

im going to check my CPUZ

@lethal shale Should i be worried?

[lethal shale]

67 degrees

I don't think you should be worried as technically you can't go beyond 100% as the internet says

But having OCD I would worry lol

[torn bluff]

sigh...

it drops sometimes to like 59% but its mostly 100%

[hot lagoon]

I heard bve 2 it’s better than it’s other one, but I wanna know what’s the difference between those two if anyone can tell me.

[lethal shale]

Let me send you something

[torn bluff]

Sure

[proud elbow]

sorry not told you two before, but the nan issue actually causes the model not learn anything at all, so it won't sound like the dataset at all.
so the only to do is use another fork like latest mainline RVC or Applio, which work on the same gpu ur using.

[lethal shale]

Most comments are like that

[torn bluff]

THANK GOD!

[lethal shale]

Ah you gotta be kidding

fuck

[torn bluff]

Wait so what you are tell me is... ive wasted my time... and my model won't even work....??

oh my...

[proud elbow]

[torn bluff]

Bro...

[lethal shale]

I can pay 10 bucks for the model lol

[torn bluff]

So what is Applio. Please tell me that is something I can run natively and not on Google Collabs...

[lethal shale]

There is no RVC v2 there foor sure

[torn bluff]

So i might as well shut this shit down with Mangio then

[rare goblet]

Ayo? @torn bluff level 11 !!!

[lethal shale]

Wai

I will check if it works

I got 60 epooch data

[torn bluff]

You doing 110 too?

[lethal shale]

Well it will be 70 now

I will go 200 lol

[torn bluff]

lol okay. let me know then ill decide if this is fruitless or not

[lethal shale]

I am waiting for 70 now

[torn bluff]

Is Applio like Mangio or is it a Google Collabs thing?

Im not doing it now, just for future reference

Thank God. Thanks

[lethal shale]

testing time

I have zero hope

[torn bluff]

lol i would too after hearing what "japanese name" said

[lethal shale]

Always expect the worse lol

Well

[proud elbow]

too bad rn I can only do inference & colabs with an igpu laptop, though may have a dataset to work on tomorrow 💀

[lethal shale]

https://tenor.com/view/gustavo-gif-25468219

[torn bluff]

It failed

[lethal shale]

70 epooch and it sounds completely different than my data

[torn bluff]

Abort?

[lethal shale]

I just need a fkin british fem model so that I wont need to force an american accent for fuck sake

70 Epooch and it sounds horrible lol

[torn bluff]

yeah im aborting mission. ggs well played

[lethal shale]

Mission failed

I will need to scam my grandpa in american accent

[torn bluff]

i stopped training. So I should try Applio then for my GPU?

lol i have been trying for at least a week on different colabs and now this and it's all come out fruitless lol

[torn bluff]

only if the model I wanted already existed, I wouldnt have to worry about training one... 💀

[lethal shale]

The only graph I have it this

Senpai help us

[lethal shale]

SAMEEEE

[proud elbow]

yea go try it

[lethal shale]

Where do I check if there is the model I want on discord I couldn't find any

[torn bluff]

Ill have to delete Mangio first because it's a waste of space on my hard drive now 😅

[rare goblet]

Ayo? @lethal shale level 8 !!!

[lethal shale]

That guy talked some stuff but I have no single idea what he says

People speak like everyone is a master coder or something

[torn bluff]

Like @proud elbow said if i am getting nan because of my GPU i need to try one that works with it. I think they said Applio works best for me

Yeah I don't understand what it means either with changing a line code. It's out of my scope

All I know is "nan" is bad

and it's not going to work

[lethal shale]

''I don't know what I am fighting but I know that I have to fight.''

Model is crap

I can show you if you wish lol

[torn bluff]

Applio you mean?

I haven't got it yet

[sudden garden]

yall this loading thing says "478.4/0.7"

is that normal-

[lethal shale]

No the difference between my 70 epooch and other random 70 epooch

[torn bluff]

Oh you mean the model that turned out bad

Sure, show it

[proud elbow]

knew that issue since about the end of september, and finally addressed

[torn bluff]

@proud elbow How do i get Applio? I found a github for it but it's only in source code. I don't know how to use things from that, im a bit dum dum

needs lamers downloads

[lethal shale]

dmed you lol

[torn bluff]

i got it

[proud elbow]

https://discord.com/channels/1159260121998827560/1203836329172013148 included latest link in huggingface

find the Applio 3.0.8

[bronze fractal]

Whats the difference between models for UVR5?

[odd shale]

Some models on UVR got better SDR than others.

[upbeat kernel]

where do i get ov2 super pretrained model?

[open sage]

I sound weird in these vc smh how to fix that

[odd shale]

You can download it on #📰│dev-updates channel

Altho ask Poopmaster for a fixed version of those.

Also, don't use it unles you got a really short dataset.

[bronze fractal]

Thanks! trying to use UVR5 with whisperx to setup karaoke session with friends

wonder why this doesnt exist already 😭

[odd shale]

Also, you can use MVSEP

[bronze fractal]

Whats mvsep?

[odd shale]

MVSEP is a page where there are UVR models too along others that are not available on UVR

[open sage]

when i talk its repeating words

[odd shale]

If you're only interested on instrumentals, use BS Roformer

[thin stump]

nvm 8 is giving me a normal g/total graph

[odd shale]

Yea, 8 is the best for most cases

[bronze fractal]

ah i see, thanks but im mainly trying to do this locally, trying to create a work flow from youtube download > video audio seperation > vocal instrumental seperation > whisperx subtitling > then recombine the instrumental with the video and subtitle

Was mainly wondering which model is best for karaoke setting, but i literally see a model called 6_HP-Karaoke-UVR so ill stick with that for now

[open sage]

bro how to fix the voice its repeating words

[rare goblet]

Ayo? @open sage level 1 !!!

[odd shale]

Which are your PC specs?

[open sage]

rtx 4060 ti cpu amd ryzen 7 7700, MSI B650 gaming plus

[odd shale]

Maybe you didn't configure something properly.

I know someone that will help you.

@sand iris

[sand iris]

huhcat

[odd shale]

If you're happy with those settings, then everything's alright.

[sand iris]

send screenshot of your voice changer pls

[odd shale]

Sniff guy says that his model is repeating words in W-Okada

[odd shale]

Give him image perms in any case

[odd shale]

💀

You're not using W-Okada

[open sage]

what is that

[odd shale]

It's another program for voice changing.

Idk from where you did get that program you're using

[open sage]

is voice.ai

where do i download w-okada

github?

[open sage]

how to download

[odd shale]

-rt

[azure marsh]

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

[odd shale]

Check that first rentry guide.

[open sage]

what

[odd shale]

https://rentry.co/VoiceChangerGuide

I've said, check this guide.

[open sage]

I did

[hearty bobcat]

Dataset Silence removal

Trunecate wont work correctly

[torn bluff]

Thanks someone linked me one in Dms.

[thin stump]

Use audio labeling instead https://rentry.org/RVC-dataset-RX-updated scroll all the way down

[open sage]

is this a virus

[rare goblet]

Ayo? @open sage level 2 !!!

[dense wren]

I'm searching for Turk models, there are no results, where can I find it?

[rare goblet]

Ayo? @dense wren level 1 !!!

[hearty bobcat]

what version of audacity is that i just updated to latest and i dont have the label optionb

[thin stump]

Install the missing plugin. It's under the tools tab for the installer https://forum.audacityteam.org/t/how-to-download-label-sounds/90762/2

[hearty bobcat]

its not allowing me to everytime i enable it it still doesnt show even after restarting

is there any alterrnatives?

[thin stump]

See if it's disabled in your plugin manager. Effects > plugin manager. If not you have to stick with truncating silences

[hearty bobcat]

is there any presets for trunacating as mine somehow dewletes the entire track

[thin stump]

https://docs.aihub.wtf/rvc/resources/datasets/#step-2-truncating-silence . Just look extra hard for a way to get audio labeling back if this was the one you were using

[astral perch]

How can you determine your batch size?

[thin stump]

Batch size of 8 is the safest option. Sometimes your model might be suitable using a batch size of 4 (shorter datasets) instead of 8 if there's no collapse on the g/total graph

[thin stump]

Just making sure; you denoised your dataset right? Because it shouldn't delete your entire track

[astral perch]

Omg i've been using 2 this whole time

[thin stump]

oh damn, yeah not good. It's more similar to dataset but it learns less

[astral perch]

What do you mean its sounds similair but it learns less

[thin stump]

that's what SCR Elsa tested with a batch size of 1

you get a nosier gan on the graph

[astral perch]

I saw on google that people are suggesting to set the batch size to the same size as your VRAM

[sand iris]

eh

Not a good idea

[astral perch]

What do you suggest?

[sand iris]

If you overshoot it's gonna be slow as balls (unless you disabled system RAM fallback within the nvidia driver settings)

Whatever works. 6 lands you at roughly 8GB (slightly less)

[astral perch]

I have 4gb vram though

[sand iris]

Oh

I'm affraid it won't be able to fully load on that

[astral perch]

Batch size 2 is working fine I was just wondering if there was anything more optimized to use

[sand iris]

how slow is it?

[astral perch]

Getting to 500 epochs takes me approximately 24 hours

Maybe a bit less

At least 18

[sand iris]

How long is the dataset?

[astral perch]

10 minutes

Shorter ones are a bit faster

[thin stump]

Nevermind, 2 is your only option since you only have 4gb of vram. I didn't understand that you can't force it to be batch 4 or 8

[vapid gust]

ilaria rvc dont work

[half cove]

delete and disconnect runtime and run it all again

[storm bobcat]

how do i get the voice changer to play through my mic?

[vapid gust]

aigh

[half cove]

do you mean on a game or discord?

[storm bobcat]

discord

[half cove]

you need a virtual audio cable first

[storm bobcat]

i already have one but the audio from it doesnt go through the cable

[half cove]

show your settings

for discord and okada

[storm bobcat]

can i dm u the screenshot?

[sand iris]

You should be able to send it here now

[storm bobcat]

no i got it to work but ty

[rare goblet]

Ayo? @storm bobcat level 1 !!!

[clear snow]

can i use applio to make models?

[brittle wing]

can I get the output to have a British accent when the dataset does, but the destination has an American one?

[half cove]

yes

[half cove]

try turning index up when infering ig

[brittle wing]

Thanks ever so much, Poopmaster

[hearty bobcat]

is there anyway to end training early?

[hearty bobcat]

^

[half cove]

by closing cmd

[hearty bobcat]

yeah but it hasnt saved a pth file yet

its at 300 epo

[half cove]

or if ur on applio it has button

[hearty bobcat]

im using this

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

[rare goblet]

Ayo? @hearty bobcat level 2 !!!

[half cove]

ah

you can stop training then put the epoch to like 1 higher to the one you stopped at

or if you did save_small_final_model: each save checkpoint will be in the weights folder

[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]

Is it worth it to make models with .flac format dataset?

Wav / flac conserve audio quality

Wav compresses it but flac dosen't compress (i think)

[proven hill]

runtime, disconnect and delete runtime, refresh the page

[proven hill]

yes, if possible absoluely yes

[brittle wing]

but wav is best for compatibility

flac is it too? because im training on applio

[proven hill]

idk about applio

but i highly suggest you train on mainline if possible

[brittle wing]

ty for anwsering

[proven hill]

no problem

[half cove]

Both are lossess

Flac is the one with smaller file size not wav

[torn bluff]

Applio has been running one epoch per 7 minutes (only doing 100). Is that normal?

my vram is pretty low, barely 4GBs

and a Geforce GTX 1650

[rich yoke]

is rmpv always going to sound the best if time/compute is not a factor?

i am using a fork for the first time, and im used to getting really tuned results using crepe and setting the hop manually to get the best result

[silent ore]

Im not sure where to post this but I am trying to get my friend to use Realtime Voice Changer (whichever option can work for him). He's using an AMD GPU and the tutorial is a little confusing for this process. Can anyone help?

[burnt zealot]

does any1 know if i can make models on the Intel UHD 770

[lean burrow]

C:\Users\Administrator\Downloads\ApplioV3.0.8\ApplioV3.0.8\logs\LoveBada | epoch=43 | step=473 | time=12:33:16 | training_speed=0:00:06 | loss_gen_all=36.454 (decreased 10.106170654296875)

[rare goblet]

Ayo? @lean burrow level 3 !!!

[lean burrow]

C:\Users\Administrator\Downloads\ApplioV3.0.8\ApplioV3.0.8\logs\LoveBada | epoch=43 | step=473 | time=12:33:16 | training_speed=0:00:06 | loss_gen_all=36.454 (decreased 10.106170654296875)
why it too big loss_gen_all

Please help

ov2 super

[crude yoke]

anyone knows why my applio doesn't install?:CC

for some reason I can't paste images here

enefecto

[rare goblet]

Ayo? @crude yoke level 1 !!!

[lean burrow]

-local

[azure marsh]

🖥️ Local stuff

• 🍏 Applio, by IA Hispano GitHub
• Mangio-RVC-Fork, by Mangio621 Huggingface
• RVC Studio, by SayanoAI Huggingface
• AICoverGen, by SociallyIneptWeeb GitHub
• Replay, by Replay Team Website
• Original RVC, by the RVC-Project team GitHub
• GPT-SoVITS, by RVC-Boss GitHub

Need some help? 🤔

You can find more info on the #1159513888199540817 channel. If you can't find your answer, feel free to ask for help in #✨│ai-help. Credits to Faze Masta and Antasma for compiling these links.

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

[pseudo sentinel]

hi

The application does not transmit my voice immediately. My voice is heard about 5 seconds after I speak, what should I do?

[rare goblet]

Ayo? @pseudo sentinel level 1 !!!

[pseudo sentinel]

please helpppp

[pseudo sentinel]

These are my settings

what should I do

[hot lagoon]

What should I use for Separating two singers ?

[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

[spare robin]

I've been seeing a lot of drum models pop up recently. What are they for if I may ask?

[thin stump]

you play your own drum instruments/kit and it'll convert into the model's version. Idk how many brands of drums are out there but sounds neat

now if only there's fender models for guitars

[hardy oriole]

Why I can't download voicemodel

What is JASONDecodeError

[distant jungle]

hekoo

[low shard]

you are using an outdated colab

check https://docs.aihub.wtf

[forest sluice]

Not worth your time

train on colab/cloud services, or buy a pc that has a GPU capable of doing so

[hot lagoon]

Is Bve 2 out for uvr?

[sonic lagoon]

sesler nerde

tyürkçe

[open solar]

Question, but what's the recommended duration for audio clips to process via a UVR5 model? I got like hours of audio for a Vtuber that I wanna make myself since I couldn't find a good voice model but since it is superchat audio, it's like 2-3hrs long and I think that's too long to run through UVR5.

[sonic lagoon]

sesler nerde

[brittle wing]

Should I select low VRAM option for 4 GB VRAM? Is it enough? Thanks

[fallen glacier]

I'm super confused. I tried like 4 different ones but none of them worked

[spare robin]

Another question, what's a good tool, if one exists. That can help with detecting a model being overtrained?
Additionally, again if it exists. Also can detect at what point roughly it was detected being overtrained

[low shard]

did u check https://docs.aihub.wtf ?

[low shard]

Only the tensorboard, which you can see more about it in our docs https://docs.aihub.wtf

[hot lagoon]

I know Karaoke is for both back vocals and harmonies.
And Bve is great for only harmonies.

But is there a way to olny get back vocals with out any harmonies in it?

[fallen glacier]

Yes, I was reading but I download and put it where the file goes, I even watched videos. Still, it won't let me use them

[lavish hull]

Friends, how can I voice someone else's voice from text?

[low shard]

what video did u watch? Most yt tuts are old so its better to use only the odcs we provide

[low shard]

Check https://docs.aihub.wtf theres how to make rvc ai voice models by having the character voice, u cant clone a voice without having it and js having text tho lol, if you meant text to speech (tts) there is a guide for that too in the docs

[fallen glacier]

This https://www.youtube.com/watch?v=We5oYpCR3WQ And This https://www.youtube.com/watch?v=_JXbvSTGPoo&t=464s

[rare goblet]

Ayo? @fallen glacier level 1 !!!

[hot lagoon]

?

[low shard]

those are so old lol, if you wanna make an ai cover check our docs, if you want to use the voice changer check in the pinned messages of #🔍│help-w-okada

[fallen glacier]

I read through it, I basically did it but it still won't let me use them

That's what it shows

The only that works it the regular beatrice one

[lean burrow]

my loss value is very big..

please help

Last time I did 1000 epochs, it was 35 loss value

[silk hollow]

hey so whenever RVC is active it picks up on EVERY background sounds, even the echo from the headphone to the mic. Does anyone know a setting to fix this?

[pseudo sentinel]

The application does not transmit my voice immediately. My voice is heard about 5 seconds after I speak, what should I do?

pleaseeeee help me

The application does not transmit my voice immediately. My voice is heard about 5 seconds after I speak, what should I do?

[fallen glacier]

I personally would mess with the chunk, that's how I got one of them to pick up

[pseudo sentinel]

I tried but it didn't work

I will try again

[lavish hull]

Is there a YouTube tutorial?

[hot lagoon]

Can someone help?

[low shard]

Sorry but idk much of wokada, maybe you should ask in #🔍│help-w-okada or #1192011222023950368

[low shard]

Nope, just read, all those 9 months old yt tuts are outdated abd would js leave you with errors

[pastel oak]

your GPU is not good enough to run the voice changer imo

if your internet is good you can try the colab version in #🔍│help-w-okada pinned messages instead

[pseudo sentinel]

I understand, thank you very much♥♥♥

wtf

[sand iris]

?

Yeah, a hacked account managed to post an invite here... Somehow filters blocked it for all channels except here

[pseudo sentinel]

Hello, I want to play games while using the voice changing program; What kind of computer system would you recommend? 🩷

[pseudo sentinel]

thank you for deleting

[hot lagoon]

Is there a way where I can remove harmonies from back vocals? so I could get clean back vocals with no harmonies in ultimate vocal remover

[lavish hull]

Thanks for your advice

[rare goblet]

Ayo? @lavish hull level 1 !!!

[verbal oasis]

Hi guys, on weights.gg how do I get the link of the huggingface model?
https://weights.gg/

I get only links like

https://weights.gg/download?fileUrl=https://models.weights.gg/cltya78l00bpece4xdebypdng.zip

I need the huggingface link

[low shard]

You cant get the hugging face link here

Weights.gg re uploads it on their own

Are you making ai covers online?

[verbal oasis]

mmmh no no I'm using SoniTranslate, in the Custom RVC Voice you can download models from links

but apparently it does not accept weights.gg

only huggingface links

[low shard]

Which rvc tool are u using

[verbal oasis]

SoniTranslate it's a colab

https://colab.research.google.com/github/R3gm/SoniTranslate/blob/main/SoniTranslate_Colab.ipynb

[low shard]

Ohh dont know much abt that, you could download on your device the voice model from weights, then re upload it on your google drive and see if the colab supports the gdrive link

[verbal oasis]

hey nice idea

instead I can just donwload the zip voice

and re upload it on huggingface

does it have simply the index and the pth file inside?

have to try

[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

[low shard]

Maybe its better if you re upload it on your gdrive (with public access ofc) just so people on hugging face dont think you made it yk, if it doesnt work by gdrive then try upload it on hugging face but ofc put credits

[verbal oasis]

aaaah

[low shard]

rvc ai voice models have the added.index and pth yes, if they dont have those or have other files unnecessary, then i dont think it will work

[severe sand]

My tensorboard only shows data points for every 2000 steps but my models are saved every 40 steps. Does anyone know a way to make tensorboard show all data points?

[hot lagoon]

I hope someone can help me on this

[plush compass]

Everyone, when learning with mangio_crepe through Colab, what is the best crepe_hop_length?

[severe sand]

Where can I set the default values of the RVC WebUI?

[tender kite]

Is there a Google Colab or HuggingFace Space where I can use RVC models for TTS?

[low shard]

check #1212368971127590922 there is a part where it tells you the rvcs with tts

[tender kite]

Ok, thank you

[low shard]

yw

[little tulip]

How do i make a ai voice model of somebody?

[low shard]

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

[little tulip]

is there perhaps a youtube tutorial? its too complicated and shi-

[low shard]

there isnt any yt tuts, all the 9 months old tuts u see are outdated and will js give you error, just read the docs link and ask here for help if you get any problems during the process

[little tulip]

i hate readin- rahhh but i gues i gotta

[peak zephyr]

Is there a way to keep the breathing, I'm trying to have voice lines that have breathiness included, like ragged expressions and such. Is this possible right now?

[manic nebula]

Guys, in the RVC Guide, when I get the voice model to download, it gets an Error message. What could it be? I took a screenshot but I can’t send it here, I really want some help 😭😭

Where can I send my screenshot?

[odd shale]

@sand iris Give him image perms

[sand iris]

he has them ?

[manic nebula]

Okk! Check your dm

No, I got confused

Don’t check your dm

[proven hill]

old colab

use ilaria rvc

https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing

[manic nebula]

Omg thank you so much

I’ll try

[rustic sedge]

Hey, i need some help 😦
everytime i am in the colab and try to "Start" after "Installing", it ends in an error.
Now it loaded around 30 minutes to give the error again:
"traceback (most recent call last):
File "/content/Ilaria-RVC/Ilaria-RVC/app.py", line 1, in <module>
import subprocess, torch, os, traceback, sys, warnings, shutil, numpy as np
ModuleNotFoundError: No module named 'torch'"

anyone knows what to do?

Nvm it worked

[manic nebula]

Sorry, but how do I use it? I don’t understand

[plush compass]

What is the appropriate crepe hop length? Default value? (64)

[proven hill]

https://youtu.be/wh4cLMLLkj0

[proven hill]

use rmvpe

[manic nebula]

Fhanksss

[plush compass]

I have already created many rmvpe models
I'm trying mangio crepe to try something new.

[proven hill]

mangio crepe is worse than rmvpe in terms of quality, no point in using it

but if you really want, for some reason the default hop lenght is 64

[plush compass]

First of all, thank you for telling me.

[proven hill]

no prob

[plush compass]

Still, I want to try something new at least once.

[proven hill]

lol, nothing bad in trying something new

good luck!

[plush compass]

thx