RVC AI Tutorial: AI Covers Made EASY

[@aadharshaadhithya]

RVC Stands for Retrival based Voice Conversion

[@BKBLUEY]

Exactly

[@ronnenvallejo7444]

It stands for “Really Very Considerate” of you to make this video, so thanks for making it.

[@eshan3009]

THIS! This is exactly what i was looking for! Super clear tutorial, easy to understand. Shows everything step by step. GREAT TUTORIAL! And thank you very much! EDIT: I was still getting error while training. Then i realized, the path to the RVC folder shouldnt have any spaces. Mine was "I:\AI Voice Swap\Mangio-RVC-v23.7.0". The "AI Voice Swap" folder has spaces. After removing the spaces from the name, im finally able to train my own model!

[@C_lamity]

This video was so efficient, I was able to get used to this before the video was finished.

[@rockstarrecipes9477]

The ONLY video about this that I was able to understand.

[@eliramone12]

Thank you very much, I am glad I decided to watch your tutorial out of all the ones out there. This has worked for me just fine, I am now subscribed, thank you again!

[@alejandrotevez6302]

Ive been looking for this kinda stuff EVERYWHERE. YOU ARE SAINT! I LOVE YOU! I will could like this video like 50 times!

[@csansolo]

Man, your video was very detailed, very thorough, thank you, you did a good job, it was very clear.

[@HustleMillennial]

Glad it was helpful!

[@stephenkudlacz4477]

I had a different result on the first part of the instructions. When clicking on the one mentioned in the assets, I clicked on it and it only downloaded the main zip file. I went ahead and extracted the files then followed the rest of the instructions. I like this video because it's a step by step method and not a run through it fast explanation assuming you know how to install it already.

[@siavashvaezi119]

you are the best! Thank you very much. I was constantly trying different sites and searching for a week. But I always failed. Other methods, or being limited, or obsolete, or poor quality. you saved me Thank you very much my friend.

[@cameronbantleon7843]

Great videos, very helpful, no BS, youre awesome dude

[@HustleMillennial]

I appreciate that!

[@MonarchButterfly.]

​@@HustleMillennialHi could you help me to make one AI cover ? I am not able to do it...Thank you

[@swatforce3309]

THIS WORKS SOOOO WELLL THAT I'M LEAVING A SUB. As a Producer my biggest problem is the lack of rappers in my area. I and others don't like my rapping voice, but I'm quite good at imitating styles, like for instance Drake's style. I tried a Drake voice over my rapping and it's insane!!! It sounds sooo goood. This is the solution for many beat producers. If you produce beats and need a rapper but don't have one, I absolutely recommend this method. Now I wouldn't publish any songs with the AI voices, as it feels like stealing(and is), but it definitely helps adding a good voice to all my beats that have been desperately needing one. Thanks from the bottom of my heart🙏❤

[@fixelheimer3726]

maybe you can merge two voices (training with input from 2 different persons) and create a new one? then you would not be stealing 1 voice :D

[@ketankhutal511]

my go -web is not opening

[@TheDylanosaur]

After double clicking on the first download link, I never get the 7Z folder. It just closes and nothing happens. Why?

[@26nadan]

I had the same problem. If you open up winrar and look through your downloads you'll be able to see the zip file there. I wasn't able to see it in downloads at all but was able to find it through winrar where I was then able to extract it.

[@recordteamnl5721]

I'm getting this error instead of the .pth file in the weight folder: "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" Does anyone know where my .pth file is or how I can solve this?

[@IfOnly60sand70s]

A really good help to me, very nicely done tutorial! Managed to get it up and running but what I've found out after days and days of experimentation is that despite the best mics' and sound equipment and 4090 equipped PC, all of my trained voice models sound really poor... excessive hiss and noise despite having the clearest voices (with no reverb and noise), I have input into my professional Studio One Artist recording set-up. Tried different levels, compression, no compression, EQ and then no EQ... not sure what I'm doing wrong. I just wonder that this is free because it is so poor. I wonder if Eleven Labs (not free obviously) and others give a better result? Cheers

[@mrmephisto]

Could you please show how to install it on MacBook?

[@MAKAVELIVEVO]

SECOND THAT 4 MAC?

[@RADEON7900GRE]

@@MAKAVELIVEVO I know how, want me to dm u on discord? u need mac m1 tho

[@RADEON7900GRE]

@@MAKAVELIVEVO i know how

[@19Charlie65]

Yes, it works, had a fair bit of success... the only downside is the 3 voice models that I have managed to train all come out with a lot of hiss and a fair bit of noise. Tried different clear voice varieties (within the same audio clip) and different volume levels, various mics' etc, everything really. Maybe this is as good as it gets though because after all, it is free.

[@JadDeluxe]

it says runtime\python.exe not recognized as an internal command or external in the go-web command prompt and when i press any key to continue nothing pops up afterwards

[@MarianoErcole]

same for me, any solution ?

[@charliechuck3088]

Same here

[@26nadan]

I had the same problem. If you open up winrar and look through your downloads you'll be able to see the zip file there. I wasn't able to see it in downloads at all but was able to find it through winrar where I was then able to extract it.

[@C_lamity]

When it comes to moving the models, I found out that dragging them works too

[@saninamsm]

i get an error when trying to download it…

[@rb9503]

How would I resume training if it was interrupted? I did save my progress at every 15 epochs.

[@MDanielSavio]

At 13:49 press the button a nothing happends, how you fix that? I have the same.

[@MultiVerseDragonball]

Audio file has 70mb, 7:30min length , pth file has 50mb, I got a 4070ti . I see my gpu only has 30% load and im processing now over 4600seconds. Tell me this is normal someone?

[@sofiaruthana5301]

hi, is there any difference between the perfomance of this model compared to RVC WebUI

[@AmeerMPorio-r5t]

"that was not the best" BRO THST WAS SK GOODDD

[@sarsarr]

1:20 How do I uninstall that file? In which folder are the things it download? As I want to uninstall it and use the program on another computer

[@KrazyFluffyUnicorn]

Any tips on what to do when you transpose? Like if you transpose a vocal higher or lower is there any way to do that with an instrumental so that it matches? I've looked but i can't find an answer and I've attempted to change it in audacity, fl studio and even some online tools, but it sounds awful. Thanks for helping with the training, I've been trying to do that for ages 🤣

[@grandlos7278]

Very good video! Can you make a tut on how to create your own voice model, from recording to feeding it to RVC training epoches? Would be great!

[@TabbyTimeOut]

Thanks you. This was very clear and concise. Liked and subscribed.

[@Gothdir]

Very nice tutorial! It would have been nice if you show when your source audio to train has background noise, for example a you want to take Audio from a song, how to remove the noise or music. Just to have the package complete. (Its not hard to figure out tho, just at the folder and hit start)

[@MajorTomMusic]

Thank you very much for this tutorial, it's working, I had lots of fun! :))

[@berkayyzc]

Thank you so much you are the king ! I'm really searching this video !!

[@cam2life360]

Thanks for this video man!

[@atking7678]

there is no file generated in weight folder. why it happening?

[@akademieintegralnihovedomi6696]

Thank you buddy, this is VERY usefull piece.

[@dailydashvids]

The best tutorial

[@nunuarthas8680]

great tutorial great voice

[@HagaiDavidoff]

Excellent Video! Thank you so much :)

[@geraldcortez826]

I watched a few videos, and trained two models and both sounded nothing like the audio I trained them from. I trained the first with 736 epochs from 7 minutes of voice lines. and the second at 203 from five minutes. both were not even close, so I tried reinstalling and now trying to train again, going to be a while older pc with a 1650 super 4gb gpu and a i7 4790 but was told this would work even though it would be slow. I really want to train models, I followed all the steps it runs and makes the models they just suck and sound nothing like the source audio it should train them from. any ideas would be nice. thank you for any info.

[@IndigaVP]

You found error?

[@sorc1337]

IF U DONT CANT CHOOSE YOUR INDEX your audio file is to big it dont wotked for me with 11 min audio 7 min works !!!

[@Elrevisor2k]

Great 👍 what’s the minimum hardware recommended and how many samples of voice we need to train a model to clone a voice? 🙏

[@bdubz_det]

Thank you for the video. this is the best one. Question - in regard to quality of training when you mentioned family guy. Can it be off the same vocal clip or is variety important?

[@HustleMillennial]

Variety seems to be important. Not required, but definitely gives a better output.

[@Wiisa_]

i trained a model with a 25 min dataset for about 120 epochs and it didnt sound at all like the person i was trying it on i mean like not even close you couldnt even say if its a boy or a girl and i just dont know what im doing wrong

[@IndigaVP]

You found error?

[@Obi-Wan_Kenobi62]

When I try to extract the ZIP file, it says it cannot be extracted from the archive because it is an unsupported file type. is there an easy fix to make it be supported?

[@Obi-Wan_Kenobi62]

looks like all I had to do was right-click the file and hover over WinRAR and chose "extract to" instead of just "extract here"

[@mig895]

My Go-Web doesn’t work please help

[@adiencnt7948]

have you fixed it? did it say runtime python is not recognized as an internal or external command?

[@jiuthegoddess6836]

@@adiencnt7948 it says "not found"

[@JamesTheNerdd]

when training, it wont Process data, my folder is correct but nothing appears in the output

[@mikan_v2]

This tutorial looks great, it's exactly what I was looking for, so I super appreciate it. I had hoped you'd discuss epochs a little more though, if I have say twenty minutes of audio, how many epochs should I be aiming for? I see many voice models with epochs anywhere from 200-500+, but have never been sure of how to accurately judge how many I need. Also, is there a way to use this if you DON'T have an NVIDIA GPU? I really want to be able to do this but only have an Intel IRS(R) Ex Graphics card....

[@eshan3009]

-Im not an expert, but here is what i think:- -If you dont have an nvidia card, dont bother with such high epoch count. It will take VERY LONG! 200 or over 500 might be considered overtraining.- -Also, i think, if you have 20 minutes of quality audio, less epochs should do the work.- Quality > Quantity. Plz ignore my stupid comment xD

[@jonescrusher1]

@@eshan3009 Why not just delete the comment if it's stupid?

[@eshan3009]

@@jonescrusher1 So that other people reading would know that 200 epochs won't be overtraining or lesser epochs will not yield nice result.

[@jonescrusher1]

@@eshan3009 So are you saying less epochs is sufficient? I'm still not clear

[@eshan3009]

@@jonescrusher1 No. Less epochs will give you poor results. Say 50 epochs, its not enough for the AI to learn the voice. What i have seen so far is that 120 epochs: the AI starts sounding like you want it to. 250: will start giving actual good results.

[@ZevennTheKing]

Why when i train my own the results have only Index File and in the log folder?

[@jackjr.sparrow3668]

doesn't seem to work right? keeps saying connection error half way.

[@DigitizedGalaxyAlt]

the command window is gone but there's no zip file? i tried this twice with the same result

[@jaystev]

I click the .bat file but it doesn't do anything after downloading

[@MegaFer1994]

could you help me? pls, the train section "section data" does not work and i dont know why because im following your steps

[@JolieKim2795]

does it have text2voice. I can't find this feature on RVC ?

[@jaybe8877]

This is amazing, great tutorial and works so well. I was wondering if you ever found what some vocalists are missing. Many vocalists know how to sing and possibly compose songs but are missing backing tracks. If they would know how to write music or play instruments that wouldn't be a problem, but they don't. Is there an AI which can take your singing and add background music production, like guitar, piano, drums, violin, or a mix of all based on the vocal input, without the vocalist knowing how to write or play instruments? Thanks for the amazing work that you do, with all the testing and the easy explanations. Only success! ❤

[@HustleMillennial]

Right now that tech is pretty limited. I think the closest we have is an AI tool called Jukebox.

[@jamessharpe2630]

​@@HustleMillennial Theres a software called Dubler 2, turns your voice in to a music instrument. Vocal input -> midi -> DAW midi control/output It is capable of real-time conversion. USD$220 for software.

[@NickArcolla]

Bitdefender ID'd this as a threat

[@maviuki]

it doenst workk!!! is getting error

[@HimanshuSharma-mm5me]

yes you got ssolution?

[@ichwei6177]

What number should I use for "Total training epochs", my audio is 8 minutes long ?

[@soupisyummy5533]

300 or 500 epochs

[@axlpuff5017]

When I open Go-Web the prompt comes up but no web page ever comes up. Anyone else have this issue?

[@rayzoralphamimic8431]

Yes. I don't know how to fix it- have you found a solution yet?

[@roberthutchison315]

Pretty cool capability.

[@HustleMillennial]

Oh for sure!

[@begoatgreatestofalltimes733]

@@HustleMillennial Thanks a lot. On my version of Mangio RVC, I dont get the three dots or a button for downloading the audio output. What am I doing wrong? I m using Firefow latest version.

[@kekw21234]

The Infer train bat file is getting detected as a virus by Bitdefender, is that a false positive

[@Y03JR]

why do i always get an error with AI programs...........

[@kekw21234]

You need python

[@Y03JR]

​@kekw21234 i had it but nvm i fixed it somehow

[@DangSs]

when I enter my model it gives an error help me pleassee

[@Ramkumar-cr8yk]

Do i need any dependencies for this rvc ?

[@RezgamerLuc]

I cant find where it downloaded the zip file. ive tried using the search on pc option, but it never came up. any idea how i can try to locate it?

[@justinr8957]

click on the file (so you select it) then right-click on it, hover over and WinRar and click "Extract Here". Make sure its already in a folder though so it doestn get lost easily

[@RezgamerLuc]

@@justinr8957 no I can't find the zip file itself. I ran the program but it never gave me the zip file.

[@justinr8957]

@@RezgamerLuc it didn’t give me a zip file either but it did give me another file (I was still able to unzip it w the process of my previous comment) also sorry for late response

[@RezgamerLuc]

@@justinr8957 It never gave me any file other than the original one you have to run

[@dav.e4410]

@@RezgamerLuc I have the same problem, have you solved this problem yet?

[@IfOnly60sand70s]

Spoke too soon... just got this..... "The expanded size of the tensor (513) must match the existing size (1025) at non-singleton dimension" Any ideas anyone?

[@Nullify-z8y]

when I try to use my model, aswell as other made models from voice models, I get this red error message when I click convert. the cmd prompt says "AttributeError: 'NoneType' object has no attribute 'tobytes'". what do I do im completely stuck

[@Ssense717]

Easy fix and that is the text that you need to replace for audios is either missed spelled or the file is something other than wave like mp3, you may have to rename the file cause of spaces as well.

[@Nullify-z8y]

@@Ssense717 well now it just never loads. like the seconds just keep going up. and when I hit convert it says "task exception was never received" followed by "validation error for Predictbody". then I got multiple of "Field required [type=missing, input_value={'data': [], 'event_data'...on_hash': 'mkoc57kmpmi'}, input_type=dict]" messages

[@Ssense717]

@@Nullify-z8y Seems it can't find the model for the voice then. I updated my pc it started working smoothly. If not reinstall the software. If not, then maybe your pc can't run the software.

[@Nullify-z8y]

@@Ssense717 well it lets me select a model but I get the errors anyway

[@CdeeqFilmmaker]

i keep getting virus error message

[@jcunique3914]

Me too, with Bitdefender

[@Mercurial_Phantom]

@@jcunique3914 did it work?

[@xSupaire]

@@Mercurial_Phantom man down, i repeat man down xD

[@SusanEli-u5u]

Nicklaus Junctions

[@kyngbitz]

is it possible to do text 2 speech?

[@kejodayne239]

i got cuda error any idea to fix that :(((

[@endergames6446]

will this work with intel graphics?

[@jackgauci]

does this work on mac m3 chip

[@rirailab6184]

Is this working on iris xe graphic?

[@Fwuzeem]

Is this Windows only?

[@cesarsalve1512]

i allways get this error: The size of tensor a (24) must match the size of tensor b (16) at non-singleton dimension 2 I got it i allready created a model with the same name so i just added a 2 behind the name

[@bigpete2895]

Ive never ben able to get this to work. I just get infinite loading on queue. Wait nvmd after a couple of months I was able to get it to work.

[@raining114]

what about the index file? are just pth needed? can i delete them?

[@NaN_Not_a_Number]

i think the index file contains the speaker's accent. you can delete if you don't need.

[@dhirenpathak]

Thanks

[@vuchithanh]

Mình rất thích chương trình này tuy nhiền mình làm theo không được

[@wesync3088]

whats the size

[@BretRemund-p3r]

Alfreda Dale

[@Noor.alhayat]

this dosent work with us bcoz of gpu

[@suki-suki.alegendaryho]

does anyone know another ai voice converter thing?

[@keyeyeyes]

how to onmac

[@SamuelDerrick-q5o]

Herman Extension

[@IsmaelRusso-i8f]

Vernon Crest

[@Maalik.]

Does this only work on Windows? Does it work on Ubuntu (20.04 LTS)?

[@HustleMillennial]

I’m unsure to be honest.

[@KingsleyBradley-v8i]

Forrest Burg

[@1marcelfilms]

what the damn point of using a bat file to download something else.

[@betterlifeexe4378]

Thank you so much! I will be using this for a lightly animated web comic here on youtube. I will share the results with you, and maybe we can do some cross promotion if you would like! i would love that! Again, Thank you! every other release of RVC was not working for me, including remote hosting. But because this came with a deployable python env, it worked like a charm!

[@YDMCA]

minimum system or graphics requirements?

[@HustleMillennial]

Unsure of the minimum reqs, but my fiancé ran this on her waaay outdated cpu

[@YDMCA]

@@HustleMillennial can i create my own model from an auio clip

[@CedricNono-x4f]

Lillian Drives

[@BabbittRenee-o3n]

Bernier Dale

[@Sm0ogs23]

Is it working without NVIDIA gpu ?

[@HustleMillennial]

Yes

[@EdgarTag-s8h]

O'Hara Meadows

[@WhitneyCollins-q5g]

Zieme Dam

[@SpencerDave-d3x]

Strosin Plaza

[@atking7678]

can I train my voice?

[@HustleMillennial]

Yes you can!

[@DanielArcher-o5z]

Olson Path

[@FrancisMonroe-h9v]

Annabell Well