So RVC first check for F0 and if it's not it decides it's a breath/sibilance?
#✨│ai-help
quasi dagger
simple ore
this is how they are built #🔊│ai-development message
well, the generation consists of using an encoded spectrogram + f0 "helper" to produce a slice of audio
quasi dagger
Hopefully your modified training improves that as well
simple ore
I guess it is a balance of overtraining for voiced parts vs undertraining non-voiced
you either get one or the other
distant turtle
-colab
quasi dagger
I guess it is a balance of overtraining for voiced parts vs undertraining non-vo...
I think I understood it now. So RVC doesn't know if it's a S, CH or T sound, but it infers based on the white noise so it tries to produce a white noise closer to a S sound for instance if the audio has a word that contain that consonant. Correct me if I'm wrong
I'm thinking on inference part
quasi dagger
@simple ore Also I think it would be better if RVC skipped breath sounds during inference and leave the original one as it would sound more natural and you don't get much benefit of cloning a "breath sound" in my opinion
flint solar
<@155030383648440320> Also I think it would be better if RVC skipped breath soun...
If u cut all the breaths on ur dataset, they will sound robotic when inferencing
quasi dagger
If u cut all the breaths on ur dataset, they will sound robotic when inferencing
I know but I don't mean cutting it on the dataset, I mean skipping it on the inference so instead of trying to reproduce it, use the original from the input audio
That can be done manually but it's a ton of work of replacing the breaths
flint solar
That can be done manually but it's a ton of work of replacing the breaths
But yeah it’ll sound better
simple ore
I think I understood it now. So RVC doesn't know if it's a S, CH or T sound, but...
not during inference, during training
flint solar
not during inference, during training
So rvc doesn’t actually produce a S sound?
simple ore
it learns to produce S sound, since it is non-voiced sound without harmonics, the source of it is a white noise that is shifted in frequency
it takes much longer to shape such sounds into right shape comparing to voiced sounds
you can see on the screenshots above, the 'wavy' parts come into shape much sooner
simple ore
So rvc doesn’t actually produce a S sound?
what is S sound anyway?
a short burst of high frequency white noise
flint solar
what is S sound anyway?
High freq noise
simple ore
so yeah, it is just a column of noise on the spectrogram that is thru trainig shaped to match a desired sound
quasi dagger
so yeah, it is just a column of noise on the spectrogram that is thru trainig sh...
Makes a lot sense now
I've been always wondering how RVC did that
But about the breath sounds I still think it should leave the original when inferencing instead of trying to reproduce or maybe make that opt in if there's a use case where the breath sound is very different (like a robot character or something like that)
flint solar
But about the breath sounds I still think it should leave the original when infe...
So is rvc v2 complete, or is rvc v3 still a thing
I’ve been wondering
quasi dagger
So is rvc v2 complete, or is rvc v3 still a thing
Forget RVC v3 haha
flint solar
Forget RVC v3 haha
Why’s dat
simple ore
lol
quasi dagger
Original RVC developers when they said about v3 they were thinking of implementing it using vocos instead of hifigans but I think it didn't turn out well and also they gave up on the project. But @simple ore is working on improving the current RVC, that's the far we're gonna get I assume
simple ore
we are dabbling with different discriminator replacements to make it possible to train better models, but it requires a creation of a pretrain
which is expensive
also a better model takes way too much vram making it impossible for most people to use it
flint solar
we are dabbling with different discriminator replacements to make it possible to...
I have a question
Would replacing Hubert make any difference?
simple ore
audio needs to be coded.. it does not need to be hubert
just some other quantization method will do
terse juniper
yo where can i find the download link
flint solar
yo where can i find the download link
For what
flint solar
for rvc app
-rvc
AI HUB Docs
broken summit
hey, im pretty new to RVC and stuff, and im trying to make a voice clone of myself, and I need assistance training custom model of my own. im using Applio v3.7.2. I put in the directory of the samples i have and it says i have preprocessed the dataset sucessfuly, pitch extracted sucessfully, but when i try to use the training thing it says it completed but the terminal window says it had an error. heres a screenshot:
can anyone help me??
azure marshBOT
can anyone help me??
left sky
Anyone got a good youtube tutorial or general idea on how to train my own voice, so I can hit different notes without it cutting out or freaking out
rare gobletBOT
Ayo? @left sky level 1 !!! 
zealous depot
https://huggingface.co/spaces/TheStinger/Ilaria_Audio_Analyzer doesn't work how can i find sample rate?
low shard
Anyone got a good youtube tutorial or general idea on how to train my own voice,...
There is none
There's only written guides
What's ur PC gpu
low shard
https://huggingface.co/spaces/TheStinger/Ilaria_Audio_Analyzer doesn't work how ...
U can use Spek on your PC
low shard
https://huggingface.co/spaces/TheStinger/Ilaria_Audio_Analyzer doesn't work how ...
btw let's talk here not dms
your audio sample rate is 44.1 Khz
what are you using to train?
ilaria audio analyzer doesn't work since sooooo many months
crude flame
ight
low shard
ight
u can tell PC users to use Spek
and well, idk if there's a cloud option for people who train on mobile
is there?
crude flame
i dont know of any
low shard
i dont know of any
me neither
crude flame
<@673327878288703519> u needa update https://docs.ai-hub.wtf/rvc/resources/datas...
made the change just waiting to get merged 🙂
thx for letting me know btw
low shard
thx for letting me know btw
yw
vernal lava
-rt
brittle wing
Is it okay if I use lead vocals from an inverted Acapella for the dataset?
crude flame
Is it okay if I use lead vocals from an inverted Acapella for the dataset?
wont sound the best but ye
brittle wing
wont sound the best but ye
Filtered or from inverted Acapella?
What is it's made w official instrumental
crude flame
it would be better if you used studio sessions
brittle wing
it would be better if you used studio sessions
There's no studio Acapella for the model I'm making and I know
crude flame
then what you have is fine
brittle wing
then what you have is fine
I use BSRoformer for full Acapella
crude flame
I use BSRoformer for full Acapella
duality is better but thats also fine
brittle wing
duality is better but thats also fine
It's not
modest parcel
What’s the delay if I use GeForce RTX 4060 Ti VRAM16GB?
crude flame
I have tried and it leaves out bv
dont you want it to leave out the backing vocals
so you have to use 1 less spereration model
brittle wing
dont you want it to leave out the backing vocals
It doesn't remove them fully
crude flame
then either dont use those parts or use a bv seperation model on those parts where it didnt take it out fully
brittle wing
then either dont use those parts or use a bv seperation model on those parts whe...
Nah it's a waste of time BSRoformer helps a lot in that matter
crude flame
Nah it's a waste of time BSRoformer helps a lot in that matter
okie, you do you
brittle wing
Litsa suggested that before a while also some other people
knotty moth
duality is better but thats also fine
it has bit more "fullness" but also more noise, but kim's is cleaner though the fullness is:
BS rofo < kim's < unwa's duality
knotty moth
@grizzled viper #1159290752195633273
brittle wing
keeps cutting out
like mid sentence it cuts out, then continues, then cuts out, etc
blazing geyser
I'm new, can someone please help me with a very good RVC model of a black American voice. Thanks
brittle wing
it has bit more "fullness" but also more noise, but kim's is cleaner though the ...
No BSRoformer is better and has higher SDR than Kim's Melroformer (Melroformer is suitable mostly for instrumentals) and unwa's MODELS AREN'T OFFICIAL how many times do I have to repeat that? BSRoformer is a finetuned Acapella model of best quality.
knotty moth
No BSRoformer is better and has higher SDR than Kim's Melroformer (Melroformer i...
if you mean kim's melroformer on instrumentals, it may have least vocal bleed & noise but more muddy than unwa's duality v1 and inst v1/v1e
spiral finch
how to solve this ? :
Running with the system Python.
Traceback (most recent call last):
File "C:\Users\Gaia simper\Desktop\RVC-GUI-main\rvcgui.py", line 4, in <module>
import soundfile as sf
ModuleNotFoundError: No module named 'soundfile'
Press any key to continue . . .
brittle wing
if you mean kim's melroformer on instrumentals, it may have least vocal bleed & ...
Mhm what.
But it's still an official model.
Also it can be demudded by applying highpass filter.
knotty moth
Mhm what.
and although BS roformer 2024.04/08 is one of my SOTA, it sometimes causes top line noise in the spectrogram, or when I tried normalizing DC offset to 0, it may either removes the top line or adds/enhances it
spiral finch
help :c..
brittle wing
and although BS roformer 2024.04/08 is one of my SOTA, it sometimes causes top l...
I never had problems w that one.
Mhm bsroformer is a blessing it also removes fx out of the vocals making the dataset creation easier
knotty moth
Mhm bsroformer is a blessing it also removes fx out of the vocals making the dat...
ah, I barely have issue on separating from vocal fx/vocoder in most songs older than 2010, or it might just be treated as backing vocals, unless it is adlibs 💀
modern osprey
Hello, can anyone tell me where can i find a. I voice of everglow?
simple ore
can i ask what is the cause of this issue ?
you failed to load the dataset, you failed to extract features, why you're running training ?
golden walrus
you failed to load the dataset, you failed to extract features, why you're runni...
but it said extract succesfully
simple ore
it shows 0 files
do things in the right order my dude
- preprocess, check the output, should say
- run extract features, the log should show x/x files twice
golden walrus
interesting
simple ore
with 2 hours of audio I expect ~3-4k chunks
golden walrus
then do you know why it doesn't cut properly ?
or the audio too long so it run into crash ?
golden walrus
so 1 done then 2 ?
simple ore
yes, wait until preprocess is done
golden walrus
i was too hasty then
simple ore
it is faster if your source files match the sample rate of the model
golden walrus
i don't get the 40 48k parts. cuz my file is 44k
i did make a small 30s file and used someone tool
thank you btw
dreamy seal
how do i make ai cover locally
agile lynx
how do i make ai cover locally
download applio
dreamy seal
download applio
where do i get it
agile lynx
where do i get it
from hugging face or from github
dreamy seal
from hugging face or from github
wheres the link
agile lynx
wheres the link
hallow thistle
-rvc
hallow thistle
A download link for Applio should be somewhere on its guide.
If you have a very fast GPU that's faster than Nvidia GTX 1000 in your PC, you should be able to run it. But if you don't have a very fast PC, you can run it on cloud like Colab instead or visit Weights.gg.
dreamy seal
If you have a very fast GPU that's faster than Nvidia GTX 1000 in your PC, you s...
i have a 1080 ti
rare gobletBOT
Ayo? @dreamy seal level 1 !!!
dreamy seal
is it good or nah
hallow thistle
This is a minimum, so it should be able to run it fine but not very fast.
dreamy seal
This is a minimum, so it should be able to run it fine but not very fast.
whats the time to run it
like how long does it run
hallow thistle
What are you asking about? I said it should be able to run fine. Follow the guide the bot sent here to run this program.
simple ore
why the f you're not using a compiled version on windows?
and you probably also using python 3.13 or something
simple ore
to avoid all this nonsense
like if you got the right python version, vc++ build tools, sure you can install it manually using pip install
but why
with compiled version you dont even need python installed
dreamy seal
alr
hallow thistle
Non-compiled source code of Applio is for developers who wanted to customize it for their fork, it cannot be run for general use.
simple ore
non-compiled version requires skills slightly above those the ipad generation posess
hallow thistle
The compiled Applio files should look like this.
simple ore
as long as you unzip the compiled version to a folder that is not captured by OneDrive, any troglodyte can run it
hallow thistle
And make sure you download one of these compiled ones instead of download one by one from the space.
If a compiled Applio launched successfully, it should launches your web browser for its GUI, like this.
simple ore
dreamy seal
the first one shows blank
rare gobletBOT
Ayo? @dreamy seal level 2 !!!
simple ore
oh boy
dreamy seal
so i ran it with admin
simple ore
why
dreamy seal
simple ore
and now you have a screenshot where you selected a text on a console output and by doing it froze the program
Esc
simple ore
press Esc
glacial pollen
where?
I don't hear any " cut outs "
you should upload the audio directly or a spectrogram instead of phone recorded audio
knotty moth
I'd suggest uploading audio through https://pillowcase.su/
it also has spectrogram viewer
low shard
I'd suggest uploading audio through https://pillowcase.su/
it also has spectrogr...
is this all free
yesterday i was talking with @crude flame about the docs that need to update how to see spectorgrams, replacing the not working ilaria audio analyzer with Spek
but we didn't know what to put for people who don't got a pc and train on mobile
@hot veldt it's better to not upload anything with copyrighted instrumentals here btw
knotty moth
is this all free
yea it's friendly to mobile users to see the spectrogram
low shard
yea it's friendly to mobile users to see the spectrogram
that's good
razer could add it to the docs then
the AI HUB by Weights docs 
flint solar
the AI HUB by Weights docs <:trolley:1159468147133395025>
ai hub by weights
low shard
ai hub by weights <:trolley:1159468147133395025>
AI HUB by Weights 🔥
flint solar
AI HUB by Weights 🔥
knotty moth
AI HUB by Weights 🔥
low shard
flint solar
what a conscience
low shard
lol
distant turtle
-colab
lethal barn
when i download the models and upload then into the rvc, They don't work properly and keep crashing, anyone knows why?
crude flame
I'd suggest uploading audio through https://pillowcase.su/
it also has spectrogr...
i got warning from my av that this is a sus website lol
inner jungle
does it need to be on harvest?
low shard
when i download the models and upload then into the rvc, They don't work properl...
what's ur pc gpu?
low shard
does it need to be on harvest?
harvest isn't a good pitch extraction method
it's better to use RMVPE,
What are u doin btw?
lethal barn
RX 7600
lethal barn
use some voice models
the models that come in the app works nice
but when i download one it doesnt work
inner jungle
harvest isn't a good pitch extraction method
it's better to use RMVPE,
What are ...
just trying to get ai to say/sing stuff
low shard
just trying to get ai to say/sing stuff
im guessing you're using RVC
it's better to use RMPVE yea
inner jungle
im guessing you're using RVC
it's better to use RMPVE yea
where do i download rmpve
low shard
where do i download rmpve
you should already have that
what program and tutorial did you follow?
inner jungle
i didnt
rare gobletBOT
Ayo? @inner jungle level 1 !!!
inner jungle
i just downloaded rvc
low shard
i just downloaded rvc
which? and using what tutorial?
inner jungle
which? and using what tutorial?
i didnt i just seasrched rvc and downloaded it off the github
low shard
i didnt i just seasrched rvc and downloaded it off the github
you did the setup following https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md right?
low shard
this https://github.com/Tiger14n/RVC-GUI/releases
no, thats super outdated
inner jungle
what's your pc gpu?
1650 super
low shard
1650 super
Your GPU is good enough to do inference (use models) locally (on ur pc), you won't be able to train (make models) but use them
You can:
- Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides):
- Cloud (remote good pc, easier and faster than ur PC but it's limited):
- Ilaria RVC Zero: fastest and simplest that you can get for free
- Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
- Applio Colab: max 4 hours, not granted, of GPU
It's better u don't use rvc gui tiget, is really old and never updated
low shard
what do you reccomend
if u want to do it locally: Applio
if u want to do it on cloud: they are all pretty good, maybe the easiest is weights.gg
inner jungle
i cant figure out how to download
low shard
i cant figure out how to download
you have to read the guides
abstract flame
how do i make a AI voice?
tame mica
how do i make a AI voice?
-rvc
brittle wing
Will RVC V3 ever come out?
rare gobletBOT
Ayo? @brittle wing level 18 !!!
crude flame
Will RVC V3 ever come out?
not an offical one but a community made one
brittle wing
not an offical one but a community made one
Where is it
crude flame
Where is it
not made yet but you can track progress by reading whats happening in https://discord.com/channels/1159260121998827560/1159290193619189821
twin stump
either I am an Audiophile, or the voice models on the entire internet suck.
No matter how well it is, in my voice or samples, it sounds robotic, most at the times at the end of a word.
Does anybody have the same experience? Or am I just dumb and not using some kind of filter?
brittle wing
not made yet but you can track progress by reading whats happening in https://di...
Interesting...
@crude flame I don't see anything about RVC V3 being confirmed
crude flame
<@673327878288703519> I don't see anything about RVC V3 being confirmed
again its not a offical rvc 3
here is some more info #🧬│ai-chat message
brittle wing
again its not a offical rvc 3
Hm are there any plans for V3 or another pitch extraction
crude flame
Hm are there any plans for V3 or another pitch extraction
no new pitch extraction but there are new discriminators
brittle wing
no new pitch extraction but there are new discriminators
? :0
Someone's making a fork I saw...
distant turtle
-colab
tranquil blaze
On applio is batch size considered per gpu? Like, if I have 2 gpus, should I put 4 to get batch size 8?
rare gobletBOT
Ayo? @tranquil blaze level 6 !!!
simple ore
On applio is batch size considered per gpu? Like, if I have 2 gpus, should I put...
I think it is per GPU
tranquil blaze
Okie, thank you so much :)
tranquil blaze
I just noticed I can train on fp32, my dataset is really clean, should I go for it?
simple ore
dont train with fp16, it is fast way to get garbage output
tranquil blaze
dont train with fp16, it is fast way to get garbage output
Oh I see, so should I always train on fp32?
knotty moth
I just noticed I can train on fp32, my dataset is really clean, should I go for ...
fp32 takes twice vram consumption and may be slower.
bf16 would be more efficient, which is being worked on the new codename's rvc
tranquil blaze
fp32 takes twice vram consumption and may be slower.
bf16 would be more efficie...
Ah I see I see, I think I'll stick to fp32 for now, I don't mind it being slower, since I can just be afk for the whole time, thank you :)
knotty moth
is beatrice a person/character name or the model type in the original wokada?
are you using original wokada 2.x alpha/beta? consult in #🔍│help-w-okada instead
so ur using normal rvc, I suppose?
I have no idea how that beatrice character model screws it, unless you show the error messages in the console window
rare gobletBOT
Ayo? @thorn prawn level 1 !!!
knotty moth
no you could post here
knotty moth
how about try moving the entire application to a directory path with only alphanumeric name and no spaces?
I mean like D:\MMVCServerSIO
tranquil blaze
@warm shell sorry for ping, just wanted to let you know that maybe there's a bug in cover maker, whenever I try changing the pitch of both the instrumental and the vocals, the instrumental's pitch stays the same, while the vocals' pitch changing works as intended
Also, whenever I try adding some reverb the vocals sound so low compared to when there's no reverb, could it be because it's applied after the normalization, and not before?
knotty moth
<@1032433063713575003> sorry for ping, just wanted to let you know that maybe th...
do you want the instrumentals sound like nightcore?
no it actually slows down (- pitch) or speeds up (+ pitch), then process it, then reverts the speed
usually you may want to slow down (- pitch) for the case of high-pitched female vocals or keeping the track fullband
tiny hatch
Kaka
tranquil blaze
do you want the instrumentals sound like nightcore?
no it actually slows down (-...
Yeah, but even if under the hood it works like that in practice it doesn't work at all, I just get the same instrumentals I started with when I put, say, -2 pitch
And sometimes I need a lower pitch because some male voices I work with can't go that high, or just sound bad when going that high
knotty moth
Yeah, but even if under the hood it works like that in practice it doesn't work ...
if I try to pitch down (-5) the instrumentals directly, it would sound lofi-ish or like an old school radio running out of battery lol
tranquil blaze
Well, it depends on which songs you use lol, I lowered the pitch of so many songs because of what I said above and they actually sounded better than the original :P
I mean, better than the original with the AI vocals, of course lol
Like, you can't really make a baritone guy sing a taylor swift song with the same pitch as hers, as a dumb example LMAO
knotty moth
Well, it depends on which songs you use lol, I lowered the pitch of so many song...
you can use pitch change in Audacity or most audio editors tho
tranquil blaze
I can't be bothered though LMAO
I usually make songs on the fly, when I'm out with my friends I just pull up colab and make it quick with cover maker lol
knotty moth
too bad I have no idea on the easy way for mobile users 
tranquil blaze
I can't be bothered though LMAO
And as I said above in some cases I don't have my pc on me, so it would be really bothersome to do it on my phone :P
tranquil blaze
too bad I have no idea on the easy way for mobile users <:shrug:1159570387953258...
Oh I'm actually a pc enjoyer (lol), I usually use audacity, but sometimes I can't be bothered to do everything manually lol
knotty moth
somehow -5 pitched instrumentals while having zero formant shift can sound decent, it sounds warmer and a bit old school-ish
while, +5 pitched up instrumentals may sound a little childish
tranquil blaze
somehow -5 pitched instrumentals while having zero formant shift can sound decen...
Ah I don't know shit about audio, I just know a little bit of music theory since I play the guitar, that's it hahaha :P
I don't even know what's formant shift lol
knotty moth
I don't even know what's formant shift lol
for vocals, zero formant is like kind of keeping timbre while changing pitch
tranquil blaze
Ah I see, got it, thank you :)
tranquil blaze
somehow -5 pitched instrumentals while having zero formant shift can sound decen...
Btw, is this izotope?
knotty moth
Btw, is this izotope?
yea izotope RX
tranquil blaze
Ah heck, too expensive lol
knotty moth
I just went ||ahoy||
tranquil blaze
Ah yes, I know the ||ahoy|| way myself, I just didn't trust the source even if it was one of the websites on ||fmhy||, if you know what I mean (arrrgh) lol
crude flame
Ah yes, I know the ||ahoy|| way myself, I just didn't trust the source even if i...
do you need a supplier?
grim slate
my settings r messed up atp completely i cant hear myself where i turn on a voice from voice.ai
how do i make it where i hear myself guys?
when i turn off the voice changer i can hear myself
grim slate
by uninstalling that old garbage
uhm
rare gobletBOT
Ayo? @grim slate level 1 !!!
knotty moth
alternatives?
grim slate
and this will let me imp voices from the internet?
cyan arrow
hi there
I'm fairly new to this AI voice stuff
how difficult would it be to replicate "crunchy" metal vocals?
example: https://youtu.be/fqHrXO2WBuI?t=23
knotty moth
how do i make it where i hear myself
you could consult here #🔍│help-w-okada
tranquil blaze
do you need a supplier?
For izotope I might as well, you can PM me if you want, thank you hahaha :)
low shard
you could consult here <#1159290161683767298>
real
cyan arrow
hi there
I'm fairly new to this AI voice stuff
how difficult would it be to repl...
still actively looking for the answer
rare gobletBOT
Ayo? @cyan arrow level 1 !!!
cyan arrow
shut up
tame mica
shut up
you would probably need to have a plugin/vst for making screamos
theyre p hard to make
cyan arrow
I've read something about "distorting it the right way"
uneven tartan
hi! i can't access the making ai covers with rvc # on the info channel, it says i don't have permission, can someone help me please?
azure marshBOT
hi! i can't access the making ai covers with rvc # on the info channel, it says ...
uneven tartan
!howtoask
uneven tartan
oh
olive crest
Im having a rtx 4050 laptop with intel, where should I download rvc and what version should I download? I downloaded applio one (newest version) but it seems to support AMD only
simple ore
Im having a rtx 4050 laptop with intel, where should I download rvc and what ver...
what?
it supports nvidia by default like any other AI application
you're probably missing cuda toolkit installation, get 12.1
olive crest
you're probably missing cuda toolkit installation, get 12.1
what's the differences between applio and mainline btw
simple ore
other than TTS there are some performance improvements and bugfixes
olive crest
other than TTS there are some performance improvements and bugfixes
Alright thanks for help 
rare gobletBOT
Ayo? @olive crest level 1 !!!
final bobcat
how the fuck do i use this
low shard
how the fuck do i use this
use what?
and what's ur pc gpu?
there's tons of ai programs
low shard
hi! i can't access the making ai covers with rvc # on the info channel, it says ...
yea that #1159949278270193734 still needs to be updated
what are you looking for? and what's ur pc gpu?
odd shale
how the fuck do i use this
!howtoask
odd shale
Also, what's your GPU?
copper acorn
Does anyone know why anytime I try to train a model in Applio it comes out sounding like this? https://pillowcase.su/f/bd0304ba642beef25ade5413611c3ba2
rare gobletBOT
Ayo? @copper acorn level 1 !!!
copper acorn
This is an issue I've had since the first "new" release of Applio and I decided to come back to it to see if this issue would stop but it hasn't
simple ore
is it a male or female voice?
@copper acorn
if it is a singing set, it will be really bad for normal speech
if it is a speaking set it may fail on songs with multiple voices merging
olive crest
I don't see run-install.bat in applio v3.2.7, I tried to turn off anti virus and redownload it
knotty moth
I don't see run-install.bat in applio v3.2.7, I tried to turn off anti virus and...
you dont need that in precompiled applio, just run the application
olive crest
you dont need that in precompiled applio, just run the application
it doesn't show anything, or I have to wait?
ok I found it, my folder has space
copper acorn
Is there some sort of setting I need to check for training? It was pretty straightfoward on the og rvc repository
yeah still sounds weird
simple ore
well, your set is severely lacking a content variety
copper acorn
well, your set is severely lacking a content variety
That's weird though
because I can train with RVC and not have that issue
simple ore
also try turning off process effects in preprocess
copper acorn
specifically rvc by RVC-Project
their repository I can train the same dataset
Is there a difference between how theirs and this one works that won't let me get a workign result
simple ore
here's my singing set with 12 minutes of songs, same file
simple ore
so here the pitch is a bit weird but it does clearly speak the test sequence
yours seem to miss a few sibilant consonants
and some other issues, generally that's because the dataset is missing some of the phonemes
copper acorn
I'm training off an old vocaloid, which has bad consonants so I did expect it
but still
this is the result being trained off of the same dataset, but in RVC1006
So I'm still confused as to what's causing this
knotty moth
I'm training off an old vocaloid, which has bad consonants so I did expect it
there is no good enough rvc miku model so far
copper acorn
there is no good enough rvc miku model so far
I have a V1 that actually sounds pretty good
rare gobletBOT
Ayo? @copper acorn level 2 !!!
copper acorn
but I can't train anything on applio for some reason
simple ore
yeah, try skipping process effects
copper acorn
alr
simple ore
and dont use slices > 3s
copper acorn
and dont use slices > 3s
Does everything else here look good?
crude flame
Does everything else here look good?
batch size 8 would be better
copper acorn
Huh
Well that's something I didn't know
I thought that it was the same, but just larger chunking with audio
Alr
I'll run it
knotty moth
Does everything else here look good?
use batch size 8 (or sometimes 4 for short dataset)
simple ore
I thought that it was the same, but just larger chunking with audio
just make sure your set gets sliced properly
also check 'save every weight'
copper acorn
just make sure your set gets sliced properly
How do i make sure everything gets sliced properly?
slim wadi
whats the best of these to use with a 4070ti super
rare gobletBOT
Ayo? @slim wadi level 5 !!!
simple ore
How do i make sure everything gets sliced properly?
if you're slicing it yourself, make 3s chunks, or let Applio slice one big file
copper acorn
Using the dataset creator?
simple ore
dataset creator is for colab mainly
you just point preprocess to a folder with your source .wav file
type the path and that's it
knotty moth
How do i make sure everything gets sliced properly?
there may be some singing notes longer than 3 sec, the slicer will do that anyway
slim wadi
whats the best of these to use with a 4070ti super
is there anyone that would know lol im on a new pc and everything is running so bad
copper acorn
Btw I know the instructions say to have Applio on the C: drive but could I move it to my D: drive without issue?
Cause It's just "D:\AI" That I'd be placing it in
simple ore
D is fine
what is not fine is using OneDrive folder or some weird non-english folder with spaces
simple ore
that's mostly an issue with other libraires
technically it should have no issues with utf-8 names if you enable that in regional console settings
copper acorn
Training seems to actually be closer to the time of rvc1006 which I think is a good sign
since this one used to train incredibly fast for some reason
I'm talking 500 epochs in 7 min
but that also gave me the bad result
simple ore
500 epoch is 7 minutes does not seem right
unless you forgot to slice audio and it is just training on two mute files
copper acorn
I have no idea honestly
I'll see what happens after I finish training at 300 epochs
velvet needle
Hello, I need help with the program. Every time I try to start the program I get this error. discardvirtualmemory procedure entry point not found 
I already tried everything but nothing worked for me
wet valley
it does not even open for me
rare gobletBOT
Ayo? @wet valley level 3 !!!
velvet needle
glacial pollen
<@132858642310365184> some help here (sorry for the ping i'm new here)
Just a pro tip my dude, do not ping Administration for this sort of things.
Instead, go for 'helpers' role people
And if you get no immediate response, you gotta be patient
wet valley
Just a pro tip my dude, do not ping Administration for this sort of things.
Inst...
okay thank you for specifying
glacial pollen
app?
wet valley
the program i mean
wet valley
rvc
glacial pollen
and is it rvc or applio specifically
rancid sluice
link
wet valley
honse
glacial pollen
@simple ore Is the amd now supported on applio or not yet?
velvet needle
link
I know that ugly meme, it's from the Chilean Mapuche. Hi, I'm German. "I'm referring to the horse Juan."
rare gobletBOT
Ayo? @velvet needle level 1 !!!
copper acorn
@simple ore everything worked and sounds exactly like it should. Thank you for the help!
velvet needle
Ayo? <@1064721743639494796> level **1** !!! <:lfg:1159355870119993496>
nice
glacial pollen
so uh i just change?
Well, give me a second. I'll check
wet valley
Well, give me a second. I'll check
okay boss on you
glacial pollen
Alr, should be supported ( supposedly )
that's how it looks for the first glance
specifically ^
Also, make sure to read this:
https://github.com/IAHispano/Applio/blob/main/assets/zluda/README.md
@wet valley
wet valley
yessir
glacial pollen
Or actually, first read the readme and then decide what you gonna do, that's all I can really say
as I ain't a specialist in zluda or, well amd anymore
( ps, if you have below 6 or 8 gigs of vram, forget about training )
and if making covers / using models is your concern then you don't really need anything specific, cpu will do
wet valley
i have enough ram
glacial pollen
Alrighty then. In that case, have a read and best of luck! I gotta go back to work
✨
wet valley
wait do i need to "train" models?
glacial pollen
I mean, cause it sounded like you intended to train
that's typically when people ask about amd or mac M chips
wet valley
omg
glacial pollen
if you mean rvc or **realtime voice changer**
Because rvc =/= realtime voice changer
glacial pollen
oh man 🤦♂️ Up til now I still wonder
who the fuck spread that confusion god damn
glacial pollen
im sorry
Nah dw, whoever started the mess quite a while ago is to be pointed at lol
just for future:
realtime voice changers = W-okada or rvc's realtime voice changer ( contained in the rvc's code )
applio / rvc are for voice to voice inferencing ( inferencing means using a model, in a short, for covers, in a static way )
analog obsidian
who the fuck spread that confusion god damn
wok himself lmao
crude flame
rvc means retrieval based voice conversion
glacial pollen
tf
rare gobletBOT
Ayo? @crude flame level 32 !!!
crude flame
Ayo? <@673327878288703519> level **32** !!! <:lfg:1159355870119993496>
cool
analog obsidian
he named wokada as realtime voice changer in prior versions
velvet needle
wok himself lmao
Hi Mod, every time I try to open the program I get this error, well it's not the same but it's similar, I've already tried everything to fix it and nothing
simple ore
stop using admin account ffs
analog obsidian
Hi Mod, every time I try to open the program I get this error, well it's not the...
im not a mod
wet valley
Nah dw, whoever started the mess quite a while ago is to be pointed at lol
just...
which one you prefer?
velvet needle
im not a mod
Hello, can you help me with this?
azure marshBOT
Hello, can you help me with this?
glacial pollen
which one you prefer?
personally?
Rvc's
but that's cause I subjectively get better latency and I like to tweak more / play with more explicit settings
wet valley
is it available for amd?
analog obsidian
but that's cause I subjectively get better latency and I like to tweak more / pl...
have u tried the fork? extra is gpu accelerated there
velvet needle
idk how to fix that, maybe noobies knows
wet valley
glacial pollen
have u tried the fork? extra is gpu accelerated there
Well, I don't use the voice changers anymore ( not that I ever started seriously playing with it )
yet, wonder if it has enough tweaking options like rvc's own one does
cause like, 2 sliders ish isn't enough
analog obsidian
yet, wonder if it has enough tweaking options like rvc's own one does
has every rvc's options
glacial pollen
uuuuu
glacial pollen
and that thing is shit ( don't mind me
truth is, it doesn't work that well
and doesn't change your voice's / input's formant but the models' output ( or, well, at least in rvc's native it worked that way )
analog obsidian
also index is extremely fast there
like
i have 10% cpu usage
index 0.7 on fork
rare gobletBOT
Ayo? @analog obsidian level 55 !!!
glacial pollen
index was never an issue for me really 🤔 but then, it shouldn't really be used anyways
higher chance of artifacts
analog obsidian
index 0,7 on rvc is like 90% cpu usage for me
crude flame
index 0.7 on fork
makes barely an audible difference
analog obsidian
index was never an issue for me really 🤔 but then, it shouldn't really be used...
tru
analog obsidian
makes barely an audible difference
i know
glacial pollen
anyways, I really like amount of stuff I can tweak
doesn't have any clogging uis, just simple tweaks n go
and I'm a simple guy 👀 I see sliders, I like
glacial pollen
well
analog obsidian
the only reason i choose the fork is merely for gpu accelerated extra anyways
glacial pollen
Ooo, in that case I could try it as soon I'm free from pretrains
thanks for letting me know actually
analog obsidian
yeah no prob give it a try
crude flame
fork is also a ton better for amd
glacial pollen
that's like a year or half a year too late 😢
analog obsidian
😭
glacial pollen
now I am a proud 3060 user, so
( removed the link, people gonna try it and cry over the bugs lol - jk, don't take it serious new people. I'm friendly )
crude flame
( removed the link, people gonna try it and cry over the bugs lol - jk, don't ta...
ight omw to complain
analog obsidian
best thing about the fork is that debloated every unnecesary shit wokada has like checking if huggingface is up and downloading those cringe anime voices
glacial pollen
lmao
glacial pollen
best thing about the fork is that debloated every unnecesary shit wokada has lik...
oh yea, that was annoying
glacial pollen
Yet, in far future, I plan to integrate the voice changer directly into the ui / rvc
just a concept but eh, we'll see
crude flame
Yet, in far future, I plan to integrate the voice changer directly into the ui /...
that would be nice
glacial pollen
so u can use it even if huggingface is ded
analog obsidian
<:nails:1159569314848972891>
i still dont know why wok added that like WHY 😭
glacial pollen
i still dont know why wok added that like WHY 😭
good question 🙃
tho ye, I gotta go for now. Still need to figure out all-speakers-at-once processing
dire gulch
I downloaded this RVC. This is like from a year ago. Does it still work?
https://github.com/Mangio621/Mangio-RVC-Fork/releases
analog obsidian
I downloaded this RVC. This is like from a year ago. Does it still work?
https:/...
no
analog obsidian
Can I get a download link to a RVC that does work?
u have two options
applio, a fork of rvc that is currently getting quality of life updates
original rvc, the... original rvc 
both give the same results
velvet needle
Hi, does anyone know how to fix this? It appears every time I try to open the program. I'm using the latest version for Windows.
analog obsidian
applio has its own applio exclusive bugs
original rvc is pretty stable but training times might be slower in some gpus compared to applio
dire gulch
I guess I'll do the original RVC.
rare gobletBOT
Ayo? @dire gulch level 1 !!!
dire gulch
Wasn't there an RVC 2?
analog obsidian
Wasn't there an RVC 2?
yes both applio and original rvc has rvc2
this is original rvc
dire gulch
Ok ty.
dire gulch
I'm having a lot of trouble trying to separate vocals from a song.
I get an error message "No such file or directory." and it's trying to target my computer's TEMP folder?
Idk what I'm doing wrong.
brittle wing
-colab
heavy kestrel
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
wtf is?
pastel oak
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not ini...
Install failed, delete the folders pretrain and model_dir , then start again
pastel oak
Idk what I'm doing wrong.
Use UVR or mvsep.com for separating vocals
heavy kestrel
Install failed, delete the folders pretrain and model_dir , then start again
а where is a model_dir
rare gobletBOT
Ayo? @heavy kestrel level 1 !!!
pastel oak
а where is a model_dir
MMVCServerSIO folder
pastel oak
Yes
heavy kestrel
young halo
does someone know how to remove background vocals with UVR?
flint solar
does someone know how to remove background vocals with UVR?
Use mel karaoke on mvsep.com
brittle wing
in case you somehow missed it
accept my friend request.
simple ore
nty
brittle wing
"GPT-SoVITS"
i have so-vits-svc-40
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
i want to sent prog screen
simple ore
i have so-vits-svc-40
can you not read?
it is a different software
so-vits is outdated as f
rocky ember
patent trellisBOT
How To Troubleshoot 
gleaming steppe
Error: 'NoneType' object has no attribute 'host_api'
knotty moth
i have so-vits-svc-40
-svc
fading lodge
can anybody help me with a voicemodel? not sure why its doing it but i need some help with it
round shuttle
yo, it keeps saying "source not found" when opening cmd
steep halo
-svc
ive never used that before but wanted to. what do people use nowadays instead of so-vits?
unique rock
Hi friends, what should I remove from my singer's voice datasets for training?
dire gulch
What's speaker/singer ID?
knotty moth
What's speaker/singer ID?
unused/unimplemented feature, you can just ignore it
dire gulch
Ok ty
silent stratus
yo is ov2 super still good with like a 7 min dataset
brittle wing
Hi friends, what should I remove from my singer's voice datasets for training?
Well some sfx voices and instrumental bleeding and voices that don't belong to the person
Who are you planning on training?
distant turtle
-colab
unkempt slate
does UVR5 (downloaded from official site, not running via Colab) comes with BS Roformer-Viper-X model? i tried refreshing the list but still couldnt find it
only found MDX23C models instead
odd shale
does UVR5 (downloaded from official site, not running via Colab) comes with BS R...
Maybe you haven't installed yet the beta roformer update that was released some time ago.
If you want i can send you an invite to the audiosep server where you can get the update.
unkempt slate
If you want i can send you an invite to the audiosep server where you can get th...
Ahh so this explains why, yeah I'd be down to try out the beta release
thick dagger
-colab
brave garnetBOT
⠀
Download for Nvidia GPUs 
Version 18a cuda
Download for AMD GPUs 
Version 18a directml
Download for Intel GPUs 
Version 18a directml
Download for Mac 
Version 17b Mac
⠀
Made with love by Ilaria 💖
sonic spire
when I click the start button in voicechanger, I get an error in the command line. returned while running Pad node. Name:'/rmvpe/mel_extractor/Pad' Status Message: CUDA error cudaErrorNoKernelImageForDevice:no kernel image is available for execution on the device
rocky kelp
I don't know which version to install, I have 2 graphics cards and I don't know which is the most powerful
jagged coral
hello guys, i have a problem, my i7 13700 is loading up to 90% by vc, while im using my 4070 in vc. who knows, how can i fix it?
fierce dagger
is there a way to train models without Nvidia GPU?
flint solar
is there a way to train models without Nvidia GPU?
u can't
crude flame
is there a way to train models without Nvidia GPU?
if you have amd yeah
intel gpus cant train at all
but if you dont have a gpu or your gpu is to bad just use a cloud service
earnest parrot
The IA weight no longer works well, the cover does not come out well 😦
fierce dagger
if you have amd yeah
I have a Gigabyte Geforce RTX 3080, i downloaded RVC and it said it was incompatible i guess
crude flame
I have a Gigabyte Geforce RTX 3080, i downloaded RVC and it said it was incompat...
prob a bug
willow gulch
what is best girl voice model
crude flame
what is best girl voice model
there is no "best" female voice model
it all depends on what you like
and your voice
willow gulch
there is no "best" female voice model
what are some reccomendations i should test
This donald trump voice changer is the best thing ive ever seen
crude flame
what are some reccomendations i should test
i dont have any recommendations
crude flame
i dont have any recommendations
just look through #1175430844685484042
knotty moth
I have a Gigabyte Geforce RTX 3080, i downloaded RVC and it said it was incompat...
make sure you are using latest rvc and try update Nvidia driver
willow gulch
how do i train my own ai voice model
flat depot
would this spec be able to run RVC locally?
oh i cant send images
my processor is amd ryzen 7 8745h, my gpu is radeon 780m graphics
simple ore
my processor is amd ryzen 7 8745h, my gpu is radeon 780m graphics
technically you can
zluda + gfx1103 libraries can run pytorch 2.3.1 cu118
peformance will be trash
flat depot
does ram amount matter? i have 24 gigs, 4 of which is dedicated to vram
simple ore
gpu uses shared memory, it is not good
you can run inference on CPU, would not make much difference vs gpu
flat depot
i see, thank you
simple ore
you can always try
applio + zluda install guide + libraries from here https://github.com/likelovewant/ROCmLibs-for-gfx1103-AMD780M-APU/releases/tag/v0.6.1.2
low shard
how do i train my own ai voice model
What's ur PC GPU
brittle wing
Can vocals with auto tune be used in a model dataset or not?
flint solar
Can vocals with auto tune be used in a model dataset or not?
yes
flint solar
Oh that won't mess up anything, right?
nop
u dont have studio vocals?
its best to use them but if they arent available its alright
knotty moth
Can vocals with auto tune be used in a model dataset or not?
as long it has enough variety, dynamic range, and consistency
brittle wing
as long it has enough variety, dynamic range, and consistency
Nice
soft palm
yo anyone know how to retrain my model
having a little trouble w/ errors
[no gpu version]
soft palm
nvm i solved it myself
limber epoch
hey! just came across Beatrice VST, which works realtime in a DAW for instant in-line vocal conversion.
this is a dream, but I'm hoping to use RVC models with this, and cannot for the life of me figure this out -- mostly because everything's in japanese.
does anyone have experience with this?
I see that there's a standalone realtime implementation of beatrice, which does have options for RVC as well as training your own models, but does that tie into the VST version? can you convert RVC to .toml that Beatrice uses? thank you !
brittle wing
-colab
brittle wing
fading lodge
I see that there's a standalone realtime implementation of beatrice, which does ...
belive you can
viscid moss
https://colab.research.google.com/github/Eddycrack864/UVR5-NO-UI/blob/main/UVR5_...
noted
azure marshBOT
CAN ANYONE HELP ME TO SETUP MY VOICE CHANGER
olive pendant
!howtoask
old kiln
knotty moth
I see that there's a standalone realtime implementation of beatrice, which does ...
ig it doesn't since RVC realtime inference (plus several load of other VSTs) is not really viable for average pc/laptop without decent GPU
viscid moss
https://colab.research.google.com/github/Eddycrack864/UVR5-NO-UI/blob/main/UVR5_...
Done
simple ore
likely yes, depends on other graphs (fm, mel)
but again the bump only looks high while in reality it is +0.4
edgy tangle
what do other charts show?
rare gobletBOT
Ayo? @edgy tangle level 5 !!!
simple ore
the usual picture with fm going to stratosphere
well, not stratospere, but up and up and up
simple ore
you can take a model saved around 8k steps
dire gulch
Can someone send me a link to the viva la vida vocal file
rare gobletBOT
Ayo? @dire gulch level 2 !!!
dire gulch
Nvm I'm just gonna download vocals the vocals off of Youtube. lol
I'm like hella new to this, but idk what an index does.
low shard
I'm like hella new to this, but idk what an index does.
It's the file containing the accent of the voice
low shard
Can someone send me a link to the viva la vida vocal file
U can also separate it urself https://docs.ai-hub.wtf/rvc/resources/vocal-isolation/
candid meteor
what is the best graph tool for seeing what is the best epoch of my model?
candid meteor
i tried tenserbord and it sucks
candid meteor
But its so hard to install and start it i tried and failed
simple ore
what
run_tb.bat:
tensorboard --logdir=C:\RVC\logs
you can skip the virtual environment completely, just install it into the global environment
candid meteor
?
candid meteor
can someone come help me in vc?
severe monolith
i have virus in new version of uvr5
knotty moth
i have virus in new version of uvr5
still trust avast?
odd shale
i have virus in new version of uvr5
Probably a false positive from your antivirus
candid meteor
i know the win defender says its a virus its not an antivirus but its safe
odd shale
i know the win defender says its a virus its not an antivirus but its safe
Then, idk but you can just disable win defender while you install UVR.
candid meteor
no,click run anyway
severe monolith
that good guys
candid meteor
can someone come help me in vc?
why when i ask a question no one answers but when you do you get all this 
keen fossil
severe monolith
why uvr5 UI don't working
candid meteor
come to vc i will help you
candid meteor
i already have python
candid meteor
simple ore
just ``python -c "code"
but again it seems that you either dont have python, did you check the box to add it to paths?
candid meteor
but again it seems that you either dont have python, did you check the box to ad...
what do you mean "add python to path"
candid meteor
i didn't check anything
simple ore
that says '[ ] Add Python to system PATH' or something like that
if you dont, then you need to do it manually
candid meteor
so i need to install python again?
simple ore
if you had it installed, then you need to edit environment variables and add it to the path
simple ore
so you have a path to python311\scripts
but not just python311
C:\Users\USER\AppData\Local\Programs\Python\Python311
candid meteor
ok i added it
patent zodiac
Try py instead of python or python3 maybe
candid meteor
simple ore
hm... maybe move python paths to the very top
you're hitting microsoft's hijack for store python
candid meteor
simple ore
where python should not list one under windows as 1st output
candid meteor
simple ore
C:\Users\USER\AppData\Local\Programs\Python\Python311\python.exe should be the 1st and only output
candid meteor
simple ore
in a new window
env variables changes do not apply to previously opened command prompts
candid meteor
can we go to a vc?
candid meteor
did you save the screen when you moved the python path up?
what do you mean
simple ore
being at the top of the list
simple ore
well, pythn works, but what the f you're doing with it I dont know
candid meteor
trying to run tensorboard
rare gobletBOT
Ayo? @candid meteor level 4 !!!
simple ore
tensorboard --logdir=c:\rvc\logs
all you need to do bro
change the path to your logs
pip install tensorboard if you have not done it
candid meteor
i have a model how do i load it
scarlet wedge
How can i make inference with kaggle without using applio notebook?
i don't like applio's inference
and i have the x2 t4 gpu
knotty moth
How can i make inference with kaggle without using applio notebook?
only inference? use this: https://huggingface.co/spaces/TheStinger/Ilaria_RVC
for the one including training (actually focused on training), use this kaggle notebook besides Applio: https://www.kaggle.com/code/hinabl/mainline
scarlet wedge
only inference? use this: https://huggingface.co/spaces/TheStinger/Ilaria_RVC
my free gpu quota in huggingface is over 😦
scarlet wedge
for the one including training (actually focused on training), use this kaggle n...
i can also make inference with this?
knotty moth
i can also make inference with this?
you can, just like applio, only with vanilla rvc features
hoary steeple
chat what do i need for the ai voice to actually work on games/discord, it still picks up my normal voice not the ai voice
candid meteor
how do i create a .bat file for tensorboard?
simple ore
if you installed it without virtual environmen, you dont need a bat file
just use it like I shown on the screenshot
candid meteor
yeah but i can put a file in the start menu folder
simple ore
okay, just make a .bat file with the very same command
in notepad and save it as .bat
scarlet wedge
for the one including training (actually focused on training), use this kaggle n...
i tried but i don't know how to upload my voice model and make inference
blazing solar
-colab
fathom crater
Traceback (most recent call last):
File "C:\Users\haris\Downloads\rvc\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\infer-web.py", line 27, in <module>
import gradio as gr
File "C:\Users\haris\Downloads\rvc\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio_init_.py", line 3, in <module>
import gradio.components as components
File "C:\Users\haris\Downloads\rvc\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 34, in <module>
from gradio_client import media_data
File "C:\Users\haris\Downloads\rvc\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio_client_init_.py", line 1, in <module>
from gradio_client.client import Client
File "C:\Users\haris\Downloads\rvc\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio_client\client.py", line 24, in <module>
from huggingface_hub import CommitOperationAdd, SpaceHardware, SpaceStage
ImportError: cannot import name 'CommitOperationAdd' from 'huggingface_hub' (C:\Users\haris\Downloads\rvc\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\huggingface_hub_init_.py)
i get this error
how do i fix this
?
simple ore
how do i fix this
stop using outdated Mangio
simple ore
apparently not enough
since the build was done over a year ago and the libraries are outdated, that's what you get
fathom crater
ah okay
issue
FileNotFoundError: [Errno 2] No such file or directory: 'pretrained/f0G40k.pth'
simple ore
and since the author did not freeze specific version installing using requirements is not gonna happen
yes, it tries to download pretrain models from huggingface
fathom crater
how do i fix this
simple ore
use RVC build or Applio
simple ore
well, good luck figuring how to update requirements
fathom crater
it worked before
rare gobletBOT
Ayo? @simple ore level 42 !!!
brittle wing
Is it a problem if some samples are flacs, others wavs?Do they all have to be the same audio format?
crude flame
Is it a problem if some samples are flacs, others wavs?Do they all have to be th...
No, they don't all have to be the same
brittle wing
No, they don't all have to be the same
What if some are WAV format and some flac?
crude flame
That's fine
brittle wing
That's fine
Despite using Mel roformer denoise 1 there's still a thin layer of noise
Should I use the mvsep aggressive one?
true ravine
zir
true ravine
iz there any indian ai voize zanzer
crude flame
Despite using Mel roformer denoise 1 there's still a thin layer of noise
RVC expects some noise so it's fine
Just make sure it's quiet
brittle wing
RVC expects some noise so it's fine
Uh but it's kinda audible
brittle wing
And melrof Denoise takes away from the vocals
crude flame
Uh but it's kinda audible
You can try training and see if RVC removes it or something
If not just be more aggressive
Or get better audio
brittle wing
You can try training and see if RVC removes it or something
Can I use KLM despite the tiny bit of noise?
crude flame
Can I use KLM despite the tiny bit of noise?
If you want
fathom crater
yo
brittle wing
If not just be more aggressive
The mvsep aggressive one?
crude flame
The mvsep aggressive one?
That works
brittle wing
That works
Uh really cause I was using that one colab
fathom crater
oh okay
crude flame
Uh really cause I was using that one colab
That one is also fine
brittle wing
crude flame
Uvr Denoiser on MVSEP?
Not the best but also works
brittle wing
Not the best but also works
How much should I set the aggressiveness to?
crude flame
How much should I set the aggressiveness to?
Leave it at default and if there is still noise increase it
brittle wing
(it adds it's own noise tho, it has been proven)
brittle wing
Leave it at default and if there is still noise increase it
Somebody said 0.5
shadow crest
tgha amd
low shard
@ivory sundial do NOT upload instrumentals for copyright reasons
knotty moth
Can I use KLM despite the tiny bit of noise?
4.1 and 4.0 may be fine but dont use 4.2 as explained in #1265083834039533588 message
marsh schooner
thought it was going slow for a 4090 does anyone know why my training not using 100%
knotty moth
thought it was going slow for a 4090 does anyone know why my training not using ...
your 4090 is too op, even using the fastest CPU won't really help
you might want to try running two instances to train each model at the same time
vale osprey
-colab
simple ore
thought it was going slow for a 4090 does anyone know why my training not using ...
4090 is something you use to train 10 hour dataset with batch 16
blazing solar
-colab
fathom whale
Is there something such as if i train a model on lets' say only 500 words and 15 minutes of audio the model would produce those 500 words better then the other English vocabulary. AND if yes could i merge 2 models trained on lets say slang and formal English so would the merged model produce both the formal and slang speech better than compared to the formal model producing slang ?
marsh schooner
4090 is something you use to train 10 hour dataset with batch 16
is the 16 batch size the best?
knotty moth
is the 16 batch size the best?
surely not for short datasets
blazing crane
anyone know what
ValueError: 32000 SR doesn't match target 40000 SR means
using Ov2 40k model for a dataset set to 40k
knotty moth
anyone know what
ValueError: 32000 SR doesn't match target 40000 SR means
you must choose the same sample rate option in the preprocess section as the pretrain
knotty moth
ok here
check if you might have had issues on preprocess and feature extraction, or perhaps completed too fast with nothing processed
knotty moth
i think nothing is wrong, idk
and you shouldn't click train index immediately after clicking start train
(or wait till the index training finished)
knotty moth
preprocess
hmm yea nothing processed at all
covert anchor
extract
covert anchor
hmm yea nothing processed at all
oh? .w.
i didn't had this problem before so, im blank, sowy
what i am doing wrong? or there is some uncompatibility issue?
im using a "flac" file btw
knotty moth
im using a "flac" file btw
I haven't tried if any formats other than wav work
knotty moth
i should export it to wav?
using audacity or another audio editor with the option to include file metadata disabled
covert anchor
i have adobe audition, that will do it
rare gobletBOT
Ayo? @covert anchor level 2 !!!
knotty moth
i have adobe audition, that will do it
like this
knotty moth
same error :c
again did you only click the train button, or both the train model and train index button immediately?