#✨│ai-help
proper shale
oh yeah... windows moment
just enable the thing to see file extensions
i forgot where it's located but i think w a google search you'd do it
torn osprey
as soon as i downloaded them i tried to drop them into this on the app
proper shale
colab, for being more reliable n free
check https://docs.aihub.wtf/ for tutorials
brittle wing
colab, for being more reliable n free
check https://docs.aihub.wtf/ for tutoria...
thanks 🙏
torn osprey
ah dang
torn osprey
i havent done this in forever so i forgot literally everything
sick narwhal
i keep getting a jsondecode error on step 2 of the easy gui, someone knows a fix? maybe cause i have a amd graphics card
proper shale
i keep getting a jsondecode error on step 2 of the easy gui, someone knows a fix...
that EasyGUI is outdated
and no colab doesn't use your gpu so dw
sick narwhal
that EasyGUI is outdated
you have a new link for me?
proper shale
prefer this (it's a guide for Ilaria RVC, one of the best new colabs)
sick narwhal
thx broo
proper shale
yw :)
brittle wing
prefer this (it's a guide for Ilaria RVC, one of the best new colabs)
https://d...
made one
rare gobletBOT
Ayo? @brittle wing level 1 !!! 
proper shale
you can send in DMs ;)
brittle wing
bet
azure marshBOT
RVC Guides (How to Make AI Cover)
oblique venture
how to cover male voice song with female voice model? whenever i try the result just a very deep voice of that female model (it sound suck)
light root
Where can I find text to rap speech?
rare gobletBOT
Ayo? @light root level 2 !!!
timid olive
how to cover male voice song with female voice model? whenever i try the result ...
you have to change the tone
timid olive
Where can I find text to rap speech?
Use suno ai to separate and clone the vocals.
light root
Use suno ai to separate and clone the vocals.
Seperate vocals from a song?
timid olive
Do you know suno ai?
light root
I want to write my own lyrics and change in to text to speech
light root
Do you know suno ai?
Nope, I've already made a rap cover but the song wasn't mine
timid olive
Okay, that's what suno ai does. But it says beta underneath.
light root
Alright thanks
dawn nebula
I want to write my own lyrics and change in to text to speech
check out gptsovits
timid olive
Alright thanks
open suno ai and enter the lyrics from the custom section, type the style, then download the song that comes out. separate the vocal and clone it
light root
open suno ai and enter the lyrics from the custom section, type the style, then ...
Thnx
light root
check out gptsovits
🫡👍
timid olive
check out gptsovits
Does it say rap?
dawn nebula
Does it say rap?
i mean if u really want to do the rap tone
light root
👍👍
tranquil cliff
in colab there is no guide yet
light root
Alright
timid olive
https://rentry.co/GPT-SoVITS-guide#/
it would be nice to have it in Turkish and a few other languages
tranquil cliff
it would be nice to have it in Turkish and a few other languages
in fact 😭
in Italian
timid olive
in fact 😭
to try to support our own language.
While they're doing it, we can, you know, maybe alphabet or something.
we'd better help
pastel oak
to try to support our own language.
While they're doing it, we can, you know, ma...
Thats something we could try to do if we have more freetime 👍
Got a few other projects for AI Hub that I want to execute first, but wouldnt mind translating a few guides already on the side
swift helm
Does this look like overtraining or should I keep it running.
proper shale
seems like it stagnated
check g/mel and g/kl to see if they're giving any improvements
swift helm
any reference as to howg/mel g/kl should look like
proper shale
any reference as to howg/mel g/kl should look like
i mean... they should be still going down
g/kl is how different model is from dataset, lower is better
g/mel is how good it is able to recreate the spectrum of the original model (or something like that), lower is better
swift helm
ah good to know 🙂 thanks. g/kl seems to be rising slowly, so lowest point is already over. g/mel is still going down
proper shale
ah good to know 🙂 thanks. g/kl seems to be rising slowly, so lowest point is al...
I'd say keep going because mel is still giving you good results
opaque imp
how to get voice?
proper shale
how to get voice?
queen escarp
what do i have to focus on when looking for a model
dawn nebula
what do i have to focus on when looking for a model
u listen to the samples, if it sounds good it will most likely osund good
queen escarp
i look for it in voice model website, they only provide the information about its epochs..etc
proper shale
i look for it in voice model website, they only provide the information about it...
yeah but you can hear samples there too
if it's weights.gg you're talking abt
queen escarp
yeah but you can hear samples there too
fr? how
pastel oak
I'd say keep going because mel is still giving you good results
Is this what you said just to determine if another OT point is possible, or to pick a model even after the stagnation even if it isnt an OT point if kl and mel are still going down?
proud elbow
what do i have to focus on when looking for a model
test in RVC and realtime okada
queen escarp
if it's weights.gg you're talking abt
oh no, it's different from the website im talking about
proper shale
Is this what you said just to determine if another OT point is possible, or to p...
it's just to see if you can keep training, so, the latter
pastel oak
since also litsa said to not pick a model after the g/total low point but just for my understanding, did you said what you said to see if another OT point is possible or not
okok
stonks
proper shale
you can also merge best mel and fm points with the best overall model (aka g/total's lowest point)
and make the best version of your model
queen escarp
test in RVC and realtime okada
welp im on mobile so it's kinda hard to switch between apps
pastel oak
you can also merge best mel and fm points with the best overall model (aka g/tot...
yoooooo
big brain, let me merge my 2000 epoch model with OT
proud elbow
welp im on mobile so it's kinda hard to switch between apps
there are some samples in weights.gg page, but still not enough to judge without listening through headphones and the audio spectrogram
proper shale
there are some samples in weights.gg page, but still not enough to judge without...
wh
that's not needed
I'd just say, listen to the samples, if it sounds good it sounds good
rare gobletBOT
Ayo? @queen escarp level 3 !!!
proper shale
but where do i find the link for the model?
well there isn't exactly a link 😭
queen escarp
well there isn't exactly a link 😭
like the huggingface link
proud elbow
I'd just say, listen to the samples, if it sounds good it sounds good
the spectrogram also reveals even some static & little noise, though yea listening it could also detect some glitches
queen escarp
oh nvm , found it
proper shale
like the huggingface link
yeah that can only be found in #1175430844685484042 i think
proper shale
the spectrogram also reveals even some static & little noise, though yea listeni...
it's like... those little noises aren't really that bad especially for realtime most of the time
quasi steppe
anyone
rare gobletBOT
Ayo? @quasi steppe level 1 !!!
queen escarp
yeah that can only be found in <#1175430844685484042> i think
they also provided it lol, you just have to click the 3 dot and it will show an option to download or copy model link etc
proper shale
how do i actually make an ai mdoel
just use this colab for training, in case you don't have a good gpu:
https://docs.aihub.wtf/rvc/cloud/rvc-disconnected/
else, you can use mainline for local training:
https://docs.aihub.wtf/rvc/local/mainline-rvc/
queen escarp
although im not sure if the link can be used to download because it's not a huggingface link
quasi steppe
just use this colab for training, in case you don't have a good gpu:
https://do...
aight
proper shale
they also provided it lol, you just have to click the 3 dot and it will show an ...
oh wow its been a while since I've used weights then haha
sudden sail
how can i use pytorch on AMD cpu and graphics card?
proud elbow
they also provided it lol, you just have to click the 3 dot and it will show an ...
it also provides the link to model post in #1175430844685484042
proper shale
how can i use pytorch on AMD cpu and graphics card?
ROCm? what will u use PyTorch for
proud elbow
though not all are posted in #1175430844685484042
sudden sail
ROCm? what will u use PyTorch for
and how do i download ROCm on Windows?
proper shale
and how do i download ROCm on Windows?
well, there isn't
it's only on linux
queen escarp
it also provides the link to model post in <#1175430844685484042>
yeah but it's very limited, i couldn't find Alice Tendou voice model in there 
sudden sail
well, there isn't
it's only on linux <:nails:1159569314848972891>
so theres no way for me to use it?
proper shale
either way, why do you need pytorch
if it's for the realtime voice changing you prolly dont need it
because on W-Okada you can convert to onnx and itll work
sudden sail
either way, why do you need pytorch <:nails:1159569314848972891>
if it's for th...
the video i watched said i needed it..
rare gobletBOT
Ayo? @sudden sail level 1 !!!
proud elbow
how can i use pytorch on AMD cpu and graphics card?
ryzen cpu and Nvidia gpu, right?
proper shale
the video i watched said i needed it..
can ya send the link here rq
proper shale
ryzen cpu and Nvidia gpu, right?
amd GPU
sudden sail
ryzen cpu and Nvidia gpu, right?
AMD GPU
sudden sail
can ya send the link here rq
alr ty
proper shale
i mean could you even train with an amd gpu
yeah no, not on windows at least
sand iris
¯_(ツ)_/¯
sand iris
rx 5k support for ROCm is wack
proper shale
rx 5k support for ROCm is wack
oh
sand iris
requires fiddling
proud elbow
ROCm on linux for sure, though RX 7000 series are more optimized on it ig
sudden sail
can ya send the link here rq
sent you the vid
sand iris
rx 400-500 : no go, unless you're highly motivated
vega series : perhaps ?
rx 5000 : fiddling required, but could work
rx 6000 : works decently (tweaking possibly needed)
rx 7000: should work OOB
ROCm chart for the time being lol
sudden sail
rx 400-500 : no go, unless you're __highly__ motivated
vega series : perhaps ?
...
how do i get ROCm tho i have an RX 7800xt
and Windows
sand iris
For this stuff ? You can't on windows yet
sudden sail
any way to bypass?
sand iris
Nope
sudden sail
okay thanks for the help my friend
quasi steppe
ugh
boreal mango
sorry for reposting if anyone sees this. can someone train my model for me to 300 epochs using local? i keep getting disconnected on colab. or can someone finish training on my account i'll send you the details of how the model was made. it's a hour plus long dataset.
odd shale
sorry for reposting if anyone sees this. can someone train my model for me to 30...
Use RVC Disconnected mate.
quasi steppe
Which colab were you using?
the disconnected version
boreal mango
Use RVC Disconnected mate.
i did. i dont have pro subscription.
odd shale
the disconnected version
experiment name normally means the name you give to the model.
quasi steppe
are numbers special characters
odd shale
i did. i dont have pro subscription.
Idk what to suggest you to do, sorry.
boreal mango
Idk what to suggest you to do, sorry.
do you use local? or just colab?
odd shale
do you use local? or just colab?
paperspace.
boreal mango
paperspace.
how fast is that? would it be fast for an hour long dataset?
odd shale
how fast is that? would it be fast for an hour long dataset?
With the pro suscription and using the A4000 GPU, an 1 hour dataset training on Mainline/Mangio should just take some hours.
But faster than colab.
boreal mango
With the pro suscription and using the A4000 GPU, an 1 hour dataset training on ...
okay.
boreal mango
With the pro suscription and using the A4000 GPU, an 1 hour dataset training on ...
could you help me train my dataset? i would be very appreciative.
odd shale
could you help me train my dataset? i would be very appreciative.
Do you want the paperspace guide?
boreal mango
Do you want the paperspace guide?
yes.
boreal mango
Do you want the paperspace guide?
but unfortunately it cost money which i dont have atm. im talking about paperspace itself
split frigate
Yo
my tensorboard
when i open it, it closes immediately
why?
odd shale
but unfortunately it cost money which i dont have atm. im talking about paperspa...
Then.. Sorry, but i can't do anything.
boreal mango
Then.. Sorry, but i can't do anything.
ok.
odd shale
Yo
my tensorboard
when i open it, it closes immediately
why?
Are you using local RVC?
split frigate
and i putted the tensorboard on the same direction with
go-web
i have and everything alright but my friend has this problem so yeah i dont know how to solve it but i will let him install Microsoft Visual C++ 14.0 just in case
quasi steppe
how long is a train for a 15 minute wav?
rare gobletBOT
Ayo? @quasi steppe level 2 !!!
low shard
<@911742715019001897> Nicky, hi buddy, can you help this guy right there?
Cant use local rvc either lol
odd shale
Cant use local rvc either lol
Me neither.
low shard
sorry
odd shale
how long is a train for a 15 minute wav?
There's no specific duration.
Even with 15 mins the training time will depend on the epoch count you wanna target to, the voice, GPU and training settings you set. Yea as Nicky said just pay attention to the Tensorboard
low shard
how long is a train for a 15 minute wav?
It depends to dataset to dataset, from gpu to gpu, so ye u cant know exactly how much it will take, even bc there could be overtraining, its just better if you use the tensorboard
odd shale
but this is alright
If it says "No dashboards" it's bc you haven't still started training anyway.
odd shale
idk how to post the screenshot
I think you can copy and paste the error.
prisma igloo
So question, What if i have a Really good, fast running computer but no GPU, is it still possible to run any RVC fork or no?
calm lotus
If my dataset sampling rate is 44, should i use 40 or 48 when training with mangio rvc?
hot lagoon
Question which one is better at separating duets bve2 or Melodyvox and why is that one better?
proud elbow
So question, What if i have a Really good, fast running computer but no GPU, is ...
even with 14900KS or threadripper alone still works, but wont outperform even budget Nvidia gpus
prisma igloo
Ok but is it Possible without one since i don't have $175 + to spend
proud elbow
used 1660 super is only around $100
odd shale
If my dataset sampling rate is 44, should i use 40 or 48 when training with mang...
Check your dataset's samplerate with Spek.
prisma igloo
Thank you, thats all i needed
glass drum
How do I use the cloned voice in Bark UI? I don't see an option to use the generatd custom voice in TTS tab
proper shale
How do I use the cloned voice in Bark UI? I don't see an option to use the gener...
the voices in here are not bark voice models
also, not sure how to use bark
glass drum
the voices in here are not bark voice models <:nails:1159569314848972891>
yeah this has nothing to do with models here, i'm asking bark ui users
proper shale
yeah this has nothing to do with models here, i'm asking bark ui users
ye i was being dumb sorry
glass drum
np lol
grizzled forum
recommended batch size/epoch for 15 mins of good quality samples?
proper shale
recommended batch size/epoch for 15 mins of good quality samples?
go for 6 or 8 ig
batch
there can't be recommended epochs for models, every model is different, so thats why we just train until we find it overtrains
hot lagoon
Can someone explane why BVE v2 is good at Separating two singers in a duet from one song?
brittle wing
!help
tawdry gullBOT
!help
LunaBot 🌙 is the perfect music bot! Feature rich with high quality music! And Custom Playlist
You can start listening music by just joinning a voice channel and typing: /play [song name or link] (Remove brackets).
We support only Spotify, soundcloud, bandcamp and more!
To view more help on a specific command or category, run
/help <command> or /help <category>
Important Links:
Support
Premium
Invite
Command Categories:
🎶: Music
💰: Premium
⚙️: Utility
📕: Admin
Select A Page From Dropdown Menu Below
LunaBot
light root
Use suno ai to separate and clone the vocals.
Hey bruv suno ai is cool but it's making it's own beat, i needed something who would convert text to rap speech a normal speech is robotic
timid olive
Hey bruv suno ai is cool but it's making it's own beat, i needed something who w...
you can separate the bottom song with the vocal remover
Try gpt-sovits
potent swan
how do i fix "index search FAILED or disabled"?
i looked it up and i tried to switch to mme but any input/output selected as mme makes the program stop responding until i close the cmd window to kill the program
light root
Try gpt-sovits
What's that?
timid olive
timid olive
What's that?
this
light root
🫡👍
worthy thicket
anybody knows how to fix this issue? i always get this error regardless what model i use
---------------------------------------------------------------------------
JSONDecodeError Traceback (most recent call last)
<ipython-input-5-134b77f94ac7> 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)
outdated colab probably
check the more recent ones
-colab
worthy thicket
thanks, will try other ones
fervent sun
That's an outdated colab
is there one you recommend
and also does that mean that my model wasn't done wrong
because i deleted it 😭😭😭😭
fervent sun
Okay thank you!
fervent sun
I deleted that model because I thought I did it wrong 😭
because of the error
So now I gotta remake it too unless I can save it
viscid moss
I deleted that model because I thought I did it wrong 😭
no way!!!
viscid moss
So now I gotta remake it too unless I can save it
old oracle
-colab
rare gobletBOT
Ayo? @old oracle level 5 !!!
fervent sun
<:AIHC_Ded:1163807516896002119>
😔😔
viscid moss
Did u delete all the files?
rare gobletBOT
Ayo? @fervent sun level 4 !!!
fervent sun
I think i can save it
viscid moss
And... the paper bin?
mellow marsh
если я скормлю для тренировки всё одним файлом, будет такой же результат?
rare gobletBOT
Ayo? @mellow marsh level 1 !!!
fervent sun
i completely forgot I uploaded it
so wait actually my error was on the GUI now that I think about it
I had the model saved
I was trying to use it to see how it sounded
Is the illaria for making voice models?
fervent sun
okay cool that's what i need to do!
azure marshBOT
RVC Guides (How to Make AI Cover)
mint veldt
Who knows the standard program for extracting audio from files from Unreal Engine 3?
swift helm
any advice on how long feature extraction should take?
I have a 60 minute dataset, rmvpe feature extraction was done in like 10 minutes.
Mangrio Crepe 64 hop feature extraction has been running for 58 minutes now 😄
brittle wing
any advice on how long feature extraction should take?
I have a 60 minute datas...
oop yeah do rmvpe not crepe
rmvpe feature extraction is higher quality, and faster
brittle wing
Who knows the standard program for extracting audio from files from Unreal Engin...
That's not exactly in our server's expertise and we can't really give advice for that 😅
mint veldt
That's not exactly in our server's expertise and we can't really give advice for...
a ok get it XD
pastel oak
set smoothing to 0.987
type in "total" in filter to display d and g total
essentially you just keep training until g/total reaches its lowest point. to determine if its reached its lowest point, youll have to keep training. If d/total is still going down after a g/total low point, as well as monitoring "mel" and "kl", then its still training and you might hit another low point of g/total
I usually keep training anothe 15k+ steps after a low point, if g/total keeps rising up then i know that one low point is the best overall model point
can you be more specific with the errors, what kind of error
dm me a screenshot if you cant put one in here
timid olive
set smoothing to 0.987
type in "total" in filter to display d and g total
esse...
Increase them? Decrease them?
proper shale
batch size and crepe hop are not related to what shad just said
he was talking about tensorboard
but eh id say increase batch
pastel oak
You can ignore crepe thats not relevant
timid olive
batch size and crepe hop are not related to what shad just said
but it was an irrelevant question.
off-topic
pastel oak
Batch size default most people use 8
i dont know the capabilities of colab though
But 8 is good
proper shale
You can ignore crepe thats not relevant
*if you're using rvmpe
proper shale
But 8 is good
8 is just ok for any 10m+ dataset
i'd test others tho
but if ur lazy
yeah just use 8
pastel oak
crepe hop length as i said you can ignore since youre using rmvpe, just any number
önemli degil, rmvpe'yi secdin ondan göre crepe hic alakasi yoktur
timid olive
önemli degil, rmvpe'yi secdin ondan göre crepe hic alakasi yoktur
türkçe biliyomuydun
Do you speak Turkish?
pastel oak
Do you speak Turkish?
i have TR in my name haha
timid olive
i have TR in my name haha
That's why I asked.
pastel oak
That's why I asked.
almanyadan türküm
timid olive
önemli degil, rmvpe'yi secdin ondan göre crepe hic alakasi yoktur
and you wrote it in Turkish.
timid olive
Türkçe konuşa bilmek için kırk fırın ekmek yemen lazım
bunu anlayabildiğin zaman Türk'sün
pastel oak
Türkçe konuşa bilmek için kırk fırın ekmek yemen lazım
😭 😂
bold mantle
i cant hear myself in okada
pastel oak
bunu anlayabildiğin zaman Türk'sün
yes yes
pastel oak
😎
ionic roost
Is there a way to make a voice sound less like the model I'm using and more like the original voice it's copying?
vestal sedge
i cant send
proper shale
check g/mel
brittle wing
check g/mel
proper shale
...that's the same image
brittle wing
...that's the same image
reef junco
how do i fix this error? https://ibb.co/c14CSFb
that link is for the image of the error btw
it wouldn't let me send the image to the discord channel
proper shale
how do i fix this error? https://ibb.co/c14CSFb
outdated colab, use https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/
its the guide for a more recent colab (ilaria rvc)
reef junco
thanks
rare gobletBOT
Ayo? @reef junco level 1 !!!
misty elk
-colab
spring orchid
proper shale
CAN SOMEBODY GIVE ME AN AI TTS
elevenlabs?
proper shale
with custom voices
gpt-sovits then
nova zealot
Does anyone know how to download a pth file in Ilaria colab after using model fusion?
proper shale
Does anyone know how to download a pth file in Ilaria colab after using model fu...
check in the weights folder, on colab's file explorer
nova zealot
check in the weights folder, on colab's file explorer
Thanks a lot, finally found it 👍🏻
rare gobletBOT
Ayo? @nova zealot level 1 !!!
azure marshBOT
RVC Guides (How to Make AI Cover)
proper shale
Thanks a lot, finally found it 👍🏻
woop, forgot to answer, yw :)
pastel oak
Theres a guide with all the links
https://docs.aihub.wtf/tts/gpt-sovits/
pastel oak
does anyone know how to use the thing?
youll have to be more specific with "the thing" 💀
pastel oak
We do not use voice ai and do not offer support for it
mild anchor
then whats this for?
pastel oak
then whats this for?
Well, retrieval based voice conversion.. including everything but voice.ai
xd
I dont have an answer for that sry 😄 if its on a colab then it would. if its local, youll have to check if theres a mac release
pastel oak
then whats this for?
If you can explain to me what you want I can give you the alternative to what we use.. which is for free
rich cove
Hi! My RVC is stuck on "Running with the runtime Python, Please wait." whenever I launch it. Please help~
rare gobletBOT
Ayo? @rich cove level 1 !!!
rich cove
Hi! My RVC is stuck on "Running with the runtime Python, Please wait." whenever ...
nvm, im stupid. i needed to press any key.
sand prawn
yo im using the google collab and when i paste the voice model i wanna use it says error what do it do ? is there a fix ?
viral schooner
Yo guys why does the colab thing give me an error when I enter the appolio thingamajig
amber orbit
how do i create models
low shard
how do i create models
Check our docs https://docs.aihub.wtf
The reason is because you are using an outdated colab
Use only the updated ones in our docs https://docs.aihub.wtf
low shard
Yw, thought u was using the applio colab at first as u talked abt it, my bad
viral schooner
all good thank you 🙂
amber olive
how do i fix the ai voice not too sound like its cracking
how to enchance unless i cant
proud elbow
how do i fix the ai voice not too sound like its cracking
- isolate, clean, denoise the input voice
- also in the settings, increase protection to reduce chance of artifacting
- unless the model quality itself
soft monolith
i legit need help idk what to do
rare gobletBOT
Ayo? @soft monolith level 1 !!!
soft monolith
my voice ai was smooth then it started to lag
terse turret
hey uhh sorry for responding a little bit late but i choose the english one or the one that says "AICoverGen-WebUI" the one made by hina
twin temple
Hey i have an error when i choose the inference for my model (that i have created) and i cannot convert
where can i send a pic of the issue?
viscid moss
hey uhh sorry for responding a little bit late but i choose the english one or t...
Is there something wrong with that Colab?
terse turret
Is there something wrong with that Colab?
no its nothing wrong
im just asking because i wasnt sure which one he was mentioning
viscid moss
where can i send a pic of the issue?
U can send the pic here, but u need lvl 1
viscid moss
im just asking because i wasnt sure which one he was mentioning
He mentioned my CoverGen Colab without UI
-colab
twin temple
U can send the pic here, but u need lvl 1 <:nails:1159569314848972891>
well okay
viscid moss
nah, this ones
twin temple
i just joined earlier
terse turret
oh alr
twin temple
U can send the pic here, but u need lvl 1 <:nails:1159569314848972891>
can i send it in dm maybe?
rare gobletBOT
Ayo? @twin temple level 1 !!!
viscid moss
can i send it in dm maybe?
sure
viscid moss
oh it works now
I see
viscid moss
it says this
Hmmm I can't see the error, but at least it works now 😅
twin temple
Hmmm I can't see the error, but at least it works now 😅
well it doesn't work i meant that now i can send picture ^^
viscid moss
I mean, where did you install it from?
barren dagger
how do I fix the rvc "connection errored out"
rare gobletBOT
Ayo? @barren dagger level 1 !!!
proud elbow
how do I fix the rvc "connection errored out"
either you accidentally closed the console window or it crashed due to some reasons like not enough free RAM
(explain the problem and what you did so I could help giving more useful solution)
icy matrix
theres a tiktoker named "timonsdreams" he makes ai cover/text to speech videos could anyone explain to me how he does it
proud elbow
have more various intonations and emotions for enough pitch range to handle
brittle wing
RVCの使い方教えて
azure marshBOT
RVC Guides (How to Make AI Cover)
twin temple
does someone know why i have this error?
runic solstice
I dont know if it's allowed, but i'm willing to pay anyone who can help me clean an accapella so i can make an AI cover out of it. I've already tried cleaning it with UVR but i still have some background/choruses on it, despite my attempts to remove them. If anyone is willing to help, please let me know, thank you so much !
twin temple
rare gobletBOT
Ayo? @twin temple level 2 !!!
small relic
ok, im new to everything and Im trying to get a ai cover of Beyonce singing Rock The Boat By Aaliyah and I need help with like everything
from start to finish
small relic
is the stuff down or not working or something?
twin temple
thx ill try this
terse turret
help..
bronze maple
I dont know if it's allowed, but i'm willing to pay anyone who can help me clean...
I'm trying to know the same things
small relic
what did you send that for?
rare gobletBOT
Ayo? @small relic level 1 !!!
small relic
what do after i download a voice model
where do i go
no, what is that?
ok so I had went to voice model channel, and found the one I wanted. What do I go to next
after i download it
im on windows
worn prawn
where the github for the updated rvc
terse turret
why wont nobody help me..
small relic
mk
where does it say download it at?
no
i thought i said that 😭
im new at this
i dont know anything
all im trying to do is convert a celebrity voice
proud elbow
im new at this
read the guide
https://docs.aihub.wtf/
twin temple
so i read the page but i don't know what i should do?
they talk about converting d and g model files to one suitable for inferance ?
how can i do that please
proud elbow
they talk about converting d and g model files to one suitable for inferance ?
neither of them, use yourmodel.pth for inference
twin temple
neither of them, use yourmodel.pth for inference
i think i don't have one or i can't find it
rare gobletBOT
Ayo? @brittle wing level 8 !!!
twin temple
i trained and followed the guide properly
do you think some files are missing for some reason?
i trained my model
proud elbow
i trained and followed the guide properly
in weights folder ig
twin temple
in weights folder ig
so for the weights folder i copied one of my D.pth files in weights i don't know if this was what i was suppose to do
yes
small relic
@brittle wing what do I do when this finally downloads
rare gobletBOT
Ayo? @small relic level 2 !!!
small relic
where the picture says download it
i heard someone say go to 'go-web' and extract it
or is that not right
nvm i did it
i had to excute it
misty turret
can someone tell where I can find a voice like packgod
proud elbow
so for the weights folder i copied one of my D.pth files in weights i don't know...
pick one of the pth files in the weights folder like yourmodel_eXXX_sXXX.pth
not the fucking D or G files
misty turret
pick one of the pth files in the weights folder like yourmodel_eXXX_sXXX.pth
not...
wait I have a question
nvm
twin temple
i think it worked but my model is so shit it makes no difference with song ill try working on it
rare gobletBOT
Ayo? @twin temple level 3 !!!
twin temple
thanks @brittle wing
quick question also i have a sample of 2 minutes to train my voice model and was wondering if it was enough also how much time do you usually train the model?
okay in my case i have a good graphics card (4070 super) so i think its fast
im trying to do a voice model of the voice of a friend of mine but i dont really have that much of content of him lol
16gb
proud elbow
4070 super 12 gb vram
twin temple
yes
proud elbow
quick question also i have a sample of 2 minutes to train my voice model and was...
it is okay, though around 10 mins dataset could potentially yield more stable and better results
brittle wing
outdated colab, use https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/
Currently, what's the most recent updated Hugginface Spaces for Local?
twin temple
it is okay, though around 10 mins dataset could potentially yield more stable an...
okay and how many time would you train? approximatively like is it in hours or minutes?
proud elbow
okay and how many time would you train? approximatively like is it in hours or m...
around 3-4 seconds per epoch for your spec ig
twin temple
i tried running it 10mn it did 212 epochs and it was very bad so im wondering if its because i haven't done it for enough time or it is the audio that is too bad
storm hatch
is there any free AI tts like elevenlabs? that is pretty good? also maybe there might be also some that support rvc's?
proud elbow
i tried running it 10mn it did 212 epochs and it was very bad so im wondering if...
also try either batch size 4 or 8, and learn more about improving the dataset here https://docs.aihub.wtf/
brittle wing
How do I get and use Tensor Board for Local?
twin temple
also try either batch size 4 or 8, and learn more about improving the dataset he...
thanks i'll try but how many epochs do you think i should i run ?
proud elbow
thanks i'll try but how many epochs do you think i should i run ?
depend on the tensorboard, you can resume training with more epochs if needed
twin temple
depend on the tensorboard, you can resume training with more epochs if needed
okay thank you i'll try monitoring it
half violet
its asking to upload an avatar what does it mean
proven sierra
Guys, maybe someone can help. I launch the program, the ping is normal, the voice works, but as soon as I alt-tab, go into the game or something else, the ping jumps in the program itself to 10-25k. The ping in the game is normal, but there is a problem in the program itself.
boreal cairn
is rvc same as w okada
pastel oak
Guys, maybe someone can help. I launch the program, the ping is normal, the voic...
Next time specify what program you use, i assume wokada
What gpu do you have
pastel oak
is rvc same as w okada
RVC has a realtime program aswell like wokada, yes
It also has another application to infer voices (making AI covers) and training voice models
boreal cairn
RVC has a realtime program aswell like wokada, yes
It also has another applicat...
for voice trolling and stuff should i use rvc or wokada
rare gobletBOT
Ayo? @boreal cairn level 2 !!!
pastel oak
for voice trolling and stuff should i use rvc or wokada
what gpu do you have
boreal cairn
rtx 3070 im pretty sure
pastel oak
rtx 3070 im pretty sure
At that point it doesnt matter which one you use
I prefer rvc because of low cpu usage and bit of higher quality, but its very manual and not as user friendly as okada
okada has simpler interface with a bit more configuration
proven sierra
Next time specify what program you use, i assume wokada
What gpu do you have
3070 Ti
pastel oak
which one you want
pastel oak
3070 Ti
Send a pic of the entire voice changer now
Yo one sec im going to eat breakfast ill help out yall in 10 mins 🙏
boreal cairn
At that point it doesnt matter which one you use
I prefer rvc because of low cp...
im gonna try out rvc and see if i have problems w it
proven sierra
Send a pic of the entire voice changer now
pastel oak
go up to 96 chunk and see if the problem still happens
you can go down to 16384 extra if you want for like 30% less cpu usage but barely a difference but if you dont like it you can keep it at that
proven sierra
go up to 96 chunk and see if the problem still happens
you can go down to 16384...
No 😦
pastel oak
No 😦
Make sure to reduce the ingame quality and limit the fps
if theres too much hardware load, youll have to keep going up in chunk to fix it which adds more delay
proven sierra
Make sure to reduce the ingame quality and limit the fps
graphics are medium, fps limit 144
blissful shuttle
my voice isnt coming when i speak
sand iris
pastel oak
my voice isnt coming when i speak
Do you have CABLE or Line 1 as virtual cable
And can you hear yourself inside the app, just not on discord/games?
blissful shuttle
And can you hear yourself inside the app, just not on discord/games?
i tried it only in the app and hear nothing except after some time some muffling...
rare gobletBOT
Ayo? @blissful shuttle level 1 !!!
pastel oak
i tried it only in the app and hear nothing except after some time some muffling...
What gpu do you have + send a screenshot of voice changer
bronze maple
What are the onnx models used for? it is better now ?
pastel oak
What are the onnx models used for? it is better now ?
onnx are a must for AMD gpu users using wokada
bronze maple
Does this reduce performance for Nvidia too?
pastel oak
Does this reduce performance for Nvidia too?
I dont know for sure, theres just no reason to use onnx for Nvidia
bronze maple
I exported an onnx file but I don't know how to import it there is no dedicated space on O-akawa
pastel oak
intel iris xe
That is not strong enough to run the voice changer
blissful shuttle
ohhh sad
pastel oak
I exported an onnx file but I don't know how to import it there is no dedicated ...
You upload the onnx file on the spot of where you uploaded the pth file. But if youre nvidia as i said dont do it
bronze maple
is my version ok locally?
pastel oak
is my version ok locally?
I dont know what kind of answer youre looking for, i guess its ok? cant tell from files
blissful shuttle
There is an online hosted alternative if your internet is decent
can you send it
my net is good
pastel oak
can you send it
round monolith
heyyy
rare gobletBOT
Ayo? @round monolith level 1 !!!
round monolith
can anyone tell me where can i find this model ? https://youtu.be/L0ubSQQeP2E?si=ekgUC7Noem6QjSeL
pastel oak
can anyone tell me where can i find this model ? https://youtu.be/L0ubSQQeP2E?si...
If its an AI voice, i dont think its public or it has a different name
blissful shuttle
https://discord.com/channels/1159260121998827560/1159290161683767298/12276249985...
does it work good?
pastel oak
does it work good?
ye
blissful shuttle
i dont need good gpu for it?
round monolith
If its an AI voice, i dont think its public or it has a different name
okay thanks !! do you know is there's a voice that sounds similar
pastel oak
i dont need good gpu for it?
You only need internet for it
blissful shuttle
ty :D
pastel oak
okay thanks !! do you know is there's a voice that sounds similar
I am not much into music so I dont know. You can search for some female singers on
https://weights.gg
https://discord.com/channels/1159260121998827560/1175430844685484042
https://voice-models.com/
round monolith
thankss
hard crystal
this may be a silly question but, has anyone found out a way to properly configure the "f0 curve file" functionality in RVC? from what i can see it replaces the f0 pitch estimation from RVC with a predefined set of pitches. but idk how to make a file for it
azure marshBOT
RVC Guides (How to Make AI Cover)
hot lagoon
Dose the Vocal Deverb Options work with out the vocal split mode?
proper shale
Currently, what's the most recent updated Hugginface Spaces for Local?
Wouldn't use Huggingface spaces for local stuff
proper shale
do I use 13k or 15k here?
latest point doesn't seem so bad, try that out
glass drum
latest point doesn't seem so bad, try that out
as in 18k?
proper shale
as in 18k?
yup
jovial zodiac
i have a question if someone could help me
everytime i start the step 2 of the google form RVC i have this :
JSONDecodeError Traceback (most recent call last)
<ipython-input-7-ebba3ce008b2> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:
5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
i don't unterstand this shit
proper shale
i have a question if someone could help me
everytime i start the step 2 of the g...
That's an outdated Colab, use stuff like Ilaria RVC (below is a guide for it):
jovial zodiac
okay i try his thanks
proven hill
That's an outdated Colab, use stuff like Ilaria RVC (below is a guide for it):
...
ilaria rvc is very based and ilaria pilled!!
proper shale
ilaria rvc is very based and ilaria pilled!!
i am indeed ilariapilled
im #1 ilaria fan
proven hill
impossible…
proper shale
impossible…
E GIRL VOICE MODELLLLLLL
proper shale
ilaria fan art spam https://media.discordapp.net/attachments/1229115868068184104...
TGAT LOOKS SO GOOD A
proven hill
TGAT LOOKS SO GOOD A
FR FR
mystic marsh
-colab
calm lotus
how do i read this tensorboard? like which one should I pick
i am using the default where it saves every 15 epoch
tough fiber
is there have difference flav or wav
pastel oak
the lowest point at ~21-22k dont know the number
calm lotus
oh okay
rare gobletBOT
Ayo? @calm lotus level 2 !!!
tranquil cliff
is there have difference flav or wav
for vocal models wav is better
flav: flav files are uncompressed audio files that use the Free Lossless Audio Codec (FLAC) format. this means that flav files are typically larger in file size than other formats, but they offer lossless audio quality, which means that the audio data is preserved exactly as it was originally recorded or produced. flav files are commonly used for high-quality audio applications.
wav: wav files, on the other hand, are also uncompressed audio files, but they use the Windows PCM (Pulse-Code Modulation) format. wav files are widely supported across various operating systems and software applications. like flav files, wav files also offer lossless audio quality, but they may be larger in file size than some other compressed audio file formats.
tough fiber
flav: flav files are uncompressed audio files that use the Free Lossless Audio C...
generally i use for okuda
so
half cove
is there have difference flav or wav
ngl it shouldnt really matter
both are lossless
just one uses lossless compression
odd shale
just one uses lossless compression
Also, RVC anyways converts flac files to .wav
half cove
yep but with datasets always use lossless to make sure you dont compress the audio further
lone lance
Guys is there a way to use voice changer with mobile?
I’ve been trying to figure out.
tranquil cliff
Guys is there a way to use voice changer with mobile?
in real time?
lone lance
any
grizzled moon
Someone drop an inference colab with gpu webui or no web ui pls thank you
smoky mantle
Hello, how can I use these sounds when talking to my friends on Discord? Is there such a method or a suitable method program?
grizzled moon
https://colab.research.google.com/drive/1mHKTGH5e3SAyDSBss1KtiYRbDdQzwSMs#scroll...
are web ui still risky? im missing on updates lately
tranquil cliff
just voice changer
https://colab.research.google.com/drive/1mHKTGH5e3SAyDSBss1KtiYRbDdQzwSMs#scrollTo=8RwuGj0LmSB_ (harder) or this https://www.weights.gg/create (easier
lone lance
https://colab.research.google.com/drive/1mHKTGH5e3SAyDSBss1KtiYRbDdQzwSMs#scroll...
okay thank you so much 👍
tranquil cliff
are web ui still risky? im missing on updates lately
this is in base64 and rot13, less risk, almost zero
tranquil cliff
Hello, how can I use these sounds when talking to my friends on Discord? Is ther...
therefore a voice change in real time
grizzled moon
this is in base64 and rot13, less risk, almost zero
Nice thanks so much
half cove
https://colab.research.google.com/drive/1mHKTGH5e3SAyDSBss1KtiYRbDdQzwSMs#scroll...
thats not realtime?
lone lance
it’s fine i just needed a voice changer
rare gobletBOT
Ayo? @lone lance level 1 !!!
lone lance
I was talking to poop
uhhh okay?
tranquil cliff
Hello, how can I use these sounds when talking to my friends on Discord? Is ther...
I don't know if this guide is still valid but I think so https://rentry.co/VoiceChangerGuide
lone lance
@tranquil cliff how do i use the harder version is there a guide i can follow?
tranquil cliff
<@954769927691198544> how do i use the harder version is there a guide i can fol...
For the difficult version there is no guide yet
I can help you
lone lance
Oh alr
lone lance
I can help you
ok
tranquil cliff
although the easy version is better
lone lance
although the easy version is better
There’s a queue
tranquil cliff
I can help you with that too
lone lance
And im pretty impatient
tranquil cliff
even with the difficult one you will have to wait a while
put T4 at the top right
in the colab
tranquil cliff
then start either base64 or rot13 as you like
lone lance
ok i did base64
tranquil cliff
did you run both the first and second base64 code?
lone lance
oh the more too?
tranquil cliff
nope
lone lance
Ok I only did base64
tranquil cliff
lower the arrow so you will see the codes
wait until the first code completes and the second one has given a blue line with a gradio code
mystic marsh
-colab
rare gobletBOT
Ayo? @mystic marsh level 1 !!!
lone lance
Do i click run?
tranquil cliff
yes
lone lance
I see 2 orange links too
tranquil cliff
let them be
lone lance
Ok i see 2 check marks that are green
1 is under it 1m and the other 0s
Uh run is 0s
rare gobletBOT
Ayo? @lone lance level 2 !!!
tranquil cliff
send scrin
tranquil cliff
yes
lone lance
lone lance
title run
tranquil cliff
below the code
lone lance
wait until the first code completes and the second one has given a blue line wit...
Yes i see that
tranquil cliff
send the scrin anyway
lone lance
how?
tranquil cliff
lone lance
lone lance
not anymore no
tranquil cliff
but do you see the links?
light root
How can I generate only the vocals in suno ai? No instrumental
lone lance
uh yeah
tranquil cliff
How can I generate only the vocals in suno ai? No instrumental
you can not
@half cove the colab is broken
half cove
what part
tranquil cliff
I think not
lone lance
I know an app that can separate the vocals and instruments
tranquil cliff
There's no other ai who can do that?
but you can remove the instrumental with programs
finite galleon
How can I generate only the vocals in suno ai? No instrumental
Try UVRing the output
finite galleon
Most of the time it doesnt get rid of the inst perfectly but its worth a shot
lone lance
There's no other ai who can do that?
I know an app that can separate the instruments
tranquil cliff
mouza
lone lance
Yes
light root
but you can remove the instrumental with programs
I just need the lyrics to be in a rap style and not robotic
rare gobletBOT
Ayo? @light root level 3 !!!
lone lance
you need to start rot13
do i click this
rare gobletBOT
Ayo? @lone lance level 3 !!!
light root
Try UVRing the output
I'll check it out 👍
tranquil cliff
do i click this
start rot13 integer
tranquil cliff
yes
tranquil cliff
do this, disconnect runtime and delete the data
lone lance
Wait I see alot of links
tranquil cliff
runtime-> disconnect runtime and delete data
lone lance
but ok I’ll delete data
tranquil cliff
and run the code
lone lance
ok
Download finished there is run left
but I see 2 links in run already
but run did not finishdownload
lone lance
??
tranquil cliff
sorry
tranquil cliff
and where do you see the links sorry
lone lance
in run at the code
lone lance
and where do you see the links sorry
it’s fine really
tranquil cliff
open the public code
tranquil cliff
download the model with the link
lone lance
I already have my own file with the person I wanna voice change as
tranquil cliff
in what voice?
lone lance
the weeknd
tranquil cliff
do you have the hf link?
lone lance
I have the file to it
tranquil cliff
you don't have to send it
lone lance
okay
tranquil cliff
put it in download model
lone lance
where?
tranquil cliff
send scrin
lone lance
rare gobletBOT
Ayo? @lone lance level 4 !!!
tranquil cliff
download voice model
lone lance
I did
tranquil cliff
now refresh
lone lance
ok
tranquil cliff
In what sense?
tranquil cliff
but if I tell you to put the link in "download voice model"
you have to put the link there 😭
lone lance
how
I don’t see download voice model but I have the model downloaded
I have the hugging face link still
tranquil cliff
lone lance
but if I tell you to put the link in "download voice model"
where is download voice model
tranquil cliff
lone lance
ohhhhhhhhhhhhhhhhhhh
tranquil cliff
you need glasses hahaha (joke)
merry lantern
what are the best settings for the voice changer?
tranquil cliff
in voice
lone lance
theweeknd
wait
what
when I just refreshed it says no interference
hold up give me a sec
tranquil cliff
tranquil cliff
put a name
tranquil cliff
press refresh the orange one (the orange one)
lone lance
ok
tranquil cliff
THE ORANGE ONE
lone lance
I did I did
tranquil cliff
is there a model?
tranquil cliff
presser
lone lance
WAIT YES I SEE
tranquil cliff
then put your audio
tranquil cliff
@round fog NOW I WANT A PROMOTION (joking, joking)
lone lance
I did
tranquil cliff
convert
tranquil cliff
back to inference
lone lance
so I click record and put my audio on?
lone lance
tranquil cliff
and so
damn oh
but how long does it take to understand how to load an AUDIO
it's literally written there
lone lance
ok
rare gobletBOT
Ayo? @lone lance level 5 !!!
lone lance
k
wow my name is purple
ok anyways
I see alot of things written audio
here all screenshot all of them
There’s 3 things written audio
1 upload file which doesn’t work
2 select file which I don’t know
3 audio input and does nothing when I click it
tranquil cliff
upload audio files
proper shale
just drag and drop the audio where it tells you to, then click on the convert button
😭
tranquil cliff
drag your file;(
timid olive
drag your file;(
iPhones can have upload problems
lone lance
exactly
timid olive
You need to do it in the files section
tranquil cliff
I didn't know it was an iPhone
lone lance
lone lance
I didn't know it was an iPhone
ipad actually
proper shale
just click on the convert button and see if it converts
thats normal on apple shit in Gradio
lone lance
okay I did click convert
tranquil cliff
I really couldn't take it anymore
lone lance
now I see alot of numbers
proper shale
thanks for the help
i need to make myself a few bucks
lone lance
I really couldn't take it anymore
my bad bro 😭
proper shale
/j
timid olive
just click on the convert button and see if it converts
can do it from colab in the files section
tranquil cliff
i need to make myself a few bucks
I won't give you any money
proper shale
my bad bro 😭
audio output should be down there
tranquil cliff
with the monthly payment right?
proper shale
I won't give you any money
truly saddening
idm honestly. keep ur money there haha
tranquil cliff
with the monthly payment right?
.
lone lance
proper shale
dont spend it on me lol
tranquil cliff
proper shale
I want a raise
You get paid????
lone lance
😭
tranquil cliff
You get paid????
I was joking bro
proper shale
Dude, come to the colab part.
no no no i think gradio screwed up
tranquil cliff
why aren't you using default ilaria rvc tho? <:shrug:1197590827527057530>
This is better, it's faster
timid olive
no no no i think gradio screwed up
It can be done in the file folder, I had a friend do it.
proper shale
This is better, it's faster
True but Mainline is still in beta and prone to issues
lone lance
why aren't you using default ilaria rvc tho? <:shrug:1197590827527057530>
what is that
is it easier?
proper shale
is it easier?
yeah it's simple enough ig
tranquil cliff
MJ dms pls
timid olive
send me the link of it
https://huggingface.co/spaces/IAHispano/Applio
Can you try this?
tranquil cliff
hf doesn't work for me on iPhone
lone lance
https://huggingface.co/spaces/IAHispano/Applio
Can you try this?
upload file don’t work
with my file
proper shale
send me the link of it
timid olive
upload file don’t work
Are you going to do a music cover?
proper shale
https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/ guide for it too
timid olive
there is colab that does it with direct youtube link
tranquil cliff
https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/ guide for it too
I have to tell you something in DMs
proper shale
there is colab that does it with direct youtube link
not needed tbh
timid olive
not needed tbh
he was already going to do tts, it won't work.
I don't know any other colab
proper shale
he was already going to do tts, it won't work.
ila rvc will work, dw
proper shale
and this girl is everywhere.
deservedly so
azure marshBOT
RVC Guides (How to Make AI Cover)
abstract sage
hey hey i installed everything and it seems to function but i cant choose my GPU it only shows cpu. how can i change that?
proven hill
and this girl is everywhere.
ilaria jumpscare
round fog
deservedly so
plush wasp
hi im used to use a .pth and a .index but now i keep seeing 2 .pth files and a config.jsob and sometimes an additional .pt file sometimes. Where do I put these files ?
dusty rampart
mystic marsh
-colab
low shard
hi im used to use a .pth and a .index but now i keep seeing 2 .pth files and a c...
where are you training the rvc model? btw you need only the namemodel.pth and added.index, also did u use tensorboard?
low shard
hey hey i installed everything and it seems to function but i cant choose my GPU...
are you using the realtime voice changer? if yes ask in #🔍│help-w-okada or #1192011222023950368 pls
low shard
but at the end I see error
what error and what are u using
lone lance
when i clicked convert 2 errors showed up
plush wasp
where are you training the rvc model? btw you need only the namemodel.pth and ad...
i did not train the model, i found it on a spreadsheet. Here's how the model looks like https://imgur.com/a/UxnZcH2
low shard
when i clicked convert 2 errors showed up
be specific on what u use and show screenshot
low shard
i did not train the model, i found it on a spreadsheet. Here's how the model loo...
where do u got that model? it doesnt look like an rvc one
lone lance
it was 2 hours ago..