Coz at the moment trying to sound excited and what not completely breaks my voice
#✨│ai-help
subtle token
subtle token
i was wondering if i could just get a dataset where it includes laughter and more excited sounds, but i think my dataset actually had some so yeah :/
spare shale
@half cove Am I saying it right? it can't be changed
spare shale
wait okay?
spare shale
brittle wing
Can I train model with free plan cpu on huggingface spaces? Does it take huge time?
spare shale
Better use Google Colab
brittle wing
Better use Google Colab
thanks
brittle wing
but may I get disconnected during training model?
rare gobletBOT
Ayo? @brittle wing level 1 !!! 
spare shale
but may I get disconnected during training model?
If you have Free colab it is possible
brittle wing
oh I see
ornate raptor
Hello bro
brittle wing
how long should it be taking to download a model? 
it's saying uploading... consistently
civic rain
No have GPU Connected
brittle wing
Thank u
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
is this ever gonna finish
i don't think so..
any way to fix that?
@dawn nebula sorry to bother, saw your name :))
dawn nebula
<@690852174267416646> sorry to bother, saw your name :))
sry i was at school
eager siren
For some reason the site while training stopped working but in cmd the epochs are still created..I just wait for it to finish and click on index?
Can anyone tell me?
rare gobletBOT
Ayo? @eager siren level 1 !!!
azure marshBOT
RVC Guides (How to Make AI Cover)
rare gobletBOT
Ayo? @brittle wing level 2 !!!
half cove
<@936673139419664414> Am I saying it right? it can't be changed
U just gotta laugh like your saying it
brittle wing
Anytime I use aicovergen no webui on colab, the voice sounds the same. @dawn nebula
I tried downloading all the files of the song, but nothing changes
sand iris
I just re-read the error you're having, and while CUDA 12 is installed, did you make sure CUBLAS is also present ?
upper furnace
but are 400 epochs ok for a minute and 17 seconds of song?
tired rapids
What's the recomendded epochs for around 10 minutes of recording?
brittle wing
If you have Free colab it is possible
thakns poyo, It worked!!
brittle wing
How do i monitor this
patent pasture
How do i monitor this
you can even just train 500 epoches and don't do brainstorm
azure marshBOT
RVC Guides (How to Make AI Cover)
stone kite
i followed the instructions of this video to a T. but where do i find a .config file for these voices? https://youtu.be/kdMpB2YCqqU?si=BlT95E1o8nyb-3o8&t=253
config.json
proper shale
i followed the instructions of this video to a T. but where do i find a .config ...
The voices in here are for RVC
That's why it doesn't have config.json
proper shale
You're welcome, let me know if ya wanna install RVC n stuff
sand iris
try adding !pip install gradio-client==0.8.1 to the first line of that code
and re-run
quartz frost
try adding `!pip install gradio-client==0.8.1` to the first line of that code
which line?
sand iris
oh you put it as last line there
quartz frost
oh you put it as last line there
I have to put it at the very top?
sand iris
goal is to have that run after the setup steps (the first few cells), and before the cell that does the conversion
sand iris
I have to put it at the very top?
yeah
quartz frost
try adding `!pip install gradio-client==0.8.1` to the first line of that code
@sand iris do I also have to use !pip that code everytime now when I convert btw?
glossy swallow
you gotta start over and dont use drive
tks
rare gobletBOT
Ayo? @glossy swallow level 3 !!!
sand iris
<@310840858767523846> do I also have to use !pip that code everytime now when I ...
if you put it once and don't re-run the setup steps, no, it'll keep it
until the notebook owner actually implements it, you will have to add it every time you re-start the notebook
quartz frost
if you put it once and don't re-run the setup steps, no, it'll keep it
alright
do I restart btw?
since i put this one on the last line
@sand iris btw tysm u saved my yt channel
i was very worried for a second
small zodiac
spare shale
U just gotta laugh like your saying it
Hahahahah
brittle wing
hii, i just trained my model through rvc disconnected
and now i have the pth and index file, and i want to make the model say another sentence instead
how do i proceed? would appreciate help
empty tusk
hello im on a 1660 when i click feature extraction this is the result? now-125,all-250,0_242.wav,(139, 768)
0_244.wav-contains nan
0_263.wav-contains nan
0_269.wav-contains nan
now-150,all-250,0_270.wav,(184, 768)
0_272.wav-contains nan
0_276.wav-contains nan
0_288.wav-contains nan
0_290.wav-contains nan
now-175,all-250,0_298.wav,(184, 768)
0_305.wav-contains nan
0_307.wav-contains nan
now-200,all-250,0_45.wav,(184, 768)
0_60.wav-contains nan
0_67.wav-contains nan
now-225,all-250,0_70.wav,(184, 768)
0_72.wav-contains nan
0_79.wav-contains nan
0_84.wav-contains nan
0_86.wav-contains nan
0_87.wav-contains nan
0_91.wav-contains nan
0_92.wav-contains nan
0_93.wav-contains nan
0_99.wav-contains nan
all-feature-done
proper shale
how do i proceed? would appreciate help
now you upload to huggingface and use a colab to inference (convert vocals)
and then just use ila rvc
https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing
eager siren
help..
proper shale
help..
it needed to be an audio file, not a folder
also
/content/MyDrive/content/dataset/EliseT would be the right path ig
proper shale
yeah i probably missed the mark on that
either way, you shouldn't be using easyGUI
imo
go for RVC disconnected
https://aihubdocs.github.io/en has a guide for it
theres the colab
eager siren
https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
is this right? index before training?
proper shale
is this right? index before training?
yes
eager siren
Is this in case I come back to continue? Should I activate now?
proper shale
Is this in case I come back to continue? Should I activate now?
this is for when u come back
so no no need to use it
not rn
hollow oak
can someone help me i created a song in the mangio rvc fork software and i created a singer's song and it told me connection error out
proper shale
can someone help me i created a song in the mangio rvc fork software and i creat...
what does cmd say
hollow oak
what does cmd say
proper shale
hmmm
i see you don't have an NVIDIA GPU
i think you'd probably be better off with colab ngl
hollow oak
i think you'd probably be better off with colab ngl
colab but it's complicated I don't understand it
twilit solstice
Can someone help me evc won't pickup my voice audio at all
hollow oak
https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sh...
Look at how long it has been
proper shale
Look at how long it has been
oh yeah thats normal for cpu
hollow oak
oh yeah thats normal for cpu
How do you know what my processor is?
Wait how long will it take to create a song bro?
proper shale
yeah because it's using the cpu
proper shale
How do you know what my processor is?
bcuz of the message displayed on cmd + long loading time
hollow oak
Wait how long will it take to create a song bro?
?
hollow oak
i honestly dont know
Can it be more than an hour?
proper shale
Can it be more than an hour?
again, i dont know
hollow oak
i honestly dont know
Oh my gosh I have to be on the site itself I can't watch YouTube or turn off the computer or just leave it like that
rare gobletBOT
Ayo? @hollow oak level 2 !!!
proper shale
Oh my gosh I have to be on the site itself I can't watch YouTube or turn off the...
I think you can see other stuff ig
lament ermine
whats top 3 RVC GUI s for windows?
proper shale
-local
azure marshBOT
proper shale
applio, mangio, original rvc
lament ermine
does applio have model training?
half cove
does applio have model training?
yes
azure marshBOT
RVC Guides (How to Make AI Cover)
eager siren
this is for when u come back
It stopped working, now just click there or do I need to turn on Dependencies again?
proper shale
It stopped working, now just click there or do I need to turn on Dependencies ag...
please tell me you exported ur model before it disconnected
if it even did disconnect
eager siren
rare gobletBOT
Ayo? @eager siren level 2 !!!
eager siren
Not yet but I haven't closed it
proper shale
u should stop training like, rn
if you have trained for longer than an hour
and export model to drive
eager siren
you probably just lost your model
💀
proper shale
ya should've stopped & exported :(
eager siren
I didn't know there was a time limit, I went to do other things while doing the epochs
😭
When I came back it was like this
proper shale
I didn't know there was a time limit, I went to do other things while doing the ...
well... now you know
eager siren
Do you recommend stopping after how many minutes?
proper shale
Do you recommend stopping after how many minutes?
like an hour or two
muted fiber
SOrry to bother guys. Trying to make my female AI voice sound baritone but she keeps singing way too high. I'm a baritone male singer
proper shale
definitely go for one hour only every day
proper shale
try setting it to 0 then
itll use the original pitch
muted fiber
oh shit. I have a laptop thats 16gb ram and no GPU. Am I able to use the newer RVC using only cpu?
proper shale
oh shit. I have a laptop thats 16gb ram and no GPU. Am I able to use the newer R...
i mean yeah
but tbh if you want to keep using it, keep using it.
id say Applio would be the best replacement
https://docs.applio.org for tutorial on how to set up ig
muted fiber
THank you!!! ❤️ ❤️ ❤️
proper shale
proper shale
THank you!!! ❤️ ❤️ ❤️
you're welcome
muted fiber
I extracted it already. I don't understand what went wrong
proper shale
I extracted it already. I don't understand what went wrong
probably a OneDrive moment
move it out of one drive
muted fiber
Hmm let me see
rare gobletBOT
Ayo? @muted fiber level 3 !!!
muted fiber
I wonder how to do that haha
true badger
hi! can someone help me please?
proper shale
Yeah sure, explain your issue and I'll try my best ere
true badger
I keep having a problem with mangio rvc
is this one
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 2.00 GiB total capacity; 1.63 GiB already allocated; 0 bytes free; 1.69 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
hot lagoon
how to fix?
ImportError: cannot import name 'set_documentation_group' from 'gradio_client.documentation' (/usr/local/lib/python3.9/dist-packages/gradio_client/documentation.py)
true badger
and cant find a solution
proper shale
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GP...
You're trying to train right?
true badger
You're trying to train right?
yes
proper shale
I've got bad news... I don't think you'll be able to train with 2GB VRAM
true badger
i have 16gb ram so is there a way to get it to vram?
proper shale
i have 16gb ram so is there a way to get it to vram?
Yeah uh, that's not how it works basically
It has to be the physical memory in the graphics card
dawn nebula
use rvc disconnected
raven iris
Im having an issue where it only picks up my voice like half the time
rare gobletBOT
Ayo? @raven iris level 1 !!!
muted fiber
@proper shale Hey so the AI you suggested, Applio, should be good for my 16gb laptop? 😄 Wont crash my stuff or too powerful right?
proper shale
<@622856015444049937> Hey so the AI you suggested, Applio, should be good for my...
Yeah it should not crash
I've seen people with 4GB RAM running the stuff
true badger
you can train on colab
I'll try!
proper shale
<:joe_what:1169303301163077672>
I got nothing sorry :(
muted fiber
I've seen people with 4GB RAM running the stuff
THank youuuuuu
muted fiber
Yeah it should not crash
so it took me to this screen
How do I import a AI Model voice onto it haha
can't seem to find the button
muted fiber
Oh ok
muted fiber
Also you can read the documentation too
I can't seem to find how to upload my downloaded file onto the Voice Model Selection Bar
proper shale
I can't seem to find how to upload my downloaded file onto the Voice Model Selec...
You can do that manually by uploading the .pth on the weights folder inside assets, and the index on the logs folder
brittle wing
i need help on the voices, i dont know what type app i need to use/download to make it work
brittle wing
well i wanna use jett voice but i dont know what app i need to download
muted fiber
Oh ok thanks I got it now
brittle wing
since i came from tiktok
proper shale
in case you wanna do realtime voice changing
https://aihubdocs.github.io/en
in case you wanna do anything else
brittle wing
https://rentry.co/VoiceChangerGuide
can i screenshare you if its okay to u
since im slow on these things
muted fiber
@proper shale Last and final question. How do I edit the pitch? It's a female AI Voice and I'm male
proper shale
<@622856015444049937> Last and final question. How do I edit the pitch? It's a f...
this is the wrong part of the GUI
u should be on the Inference section
hot lagoon
Ayo is BVE beter then 6hp or nah?
muted fiber
AH I see THANK YOU ❤️
brittle wing
which do i download
proper shale
this one yea
proper shale
which do i download
now u open start http
brittle wing
proper shale
yea
proper shale
start http or
start http
merry fiber
guys, its succesfully generating the ai voice but voice is not working ig? i cant hear anything
true badger
can someone help me by doing a model? i have the samples and all just need to train it 😢 nothing is really working for me not either colab
hot lagoon
Dose rvc convert back vocals?
distant rapids
I need help, i cant but in the voice changer for some reason, i kept trying but it will still not work...
half cove
Dose rvc convert back vocals?
if your inferring something always remove it but you can try doing it separately
half cove
I need help, i cant but in the voice changer for some reason, i kept trying but ...
give more info
half cove
you need level 1
distant rapids
how, whats level one how do i get it
half cove
just talk
distant rapids
How much\
rare gobletBOT
Ayo? @distant rapids level 1 !!!
half cove
also for voicechanger #🔍│help-w-okada
half cove
Ayo? <@1151987237018411058> level **1** !!! <:lfg:1159355870119993496>
welp there we go
hot lagoon
if your inferring something always remove it but you can try doing it separately
so is just find with the lead vocals?
jaunty yew
hi i want download an ia voice but i don't see the download in my pc can you help me pls
half cove
so is just find with the lead vocals?
you could just split the lead and backing vocals use the lead and then add em back together
hot lagoon
you could just split the lead and backing vocals use the lead and then add em ba...
so someing like this?
hot lagoon
half cove
barren acorn
well anyone know good setting for gtx 1660(6GB)?????
brave garnetBOT
⠀
Settings for Nvidia GPUs 
F0 Det.: rmvpe (suggested for all series)
RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra
Advanced Settings
Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low
⠀
Made with love by Ilaria 💖
sterile mica
where are free rvc 2?
dawn nebula
where are free rvc 2?
dawn nebula
what did u do
dawn nebula
you havent started training yet
dawn nebula
did u put it in rvc folder
molten apex
yes i did
it worked to here
then on the site it said to paste the localhost into the browser
rare gobletBOT
Ayo? @molten apex level 2 !!!
molten apex
💀
leaden thicket
Is there a verified list of AI models with huggingface links that i can view? I feel like the "voice model" section is limited
dawn nebula
Is there a verified list of AI models with huggingface links that i can view? I ...
#🔍│find-models then use / search on applio bot
or weights.gg
leaden thicket
Thank you
molten apex
bro you need to train a model first
so this?
dawn nebula
yeah
molten apex
how do i get a model in there
wide saddle
currently having an issue where i get wait web server 0
rare gobletBOT
Ayo? @wide saddle level 3 !!!
wide saddle
wait web server 1 and goes on
anyone else having the same problem
?
when i click on the start http
molten apex
how do i get a model in there
@dawn nebula
ripe gulch
my voice changer not working its not changing my voice
molten apex
train it using this?
subtle token
i can't get tensorboard to work, i download the 2 bat files from the Pin
but when i run TbL.bat i get this
??
im losing my mind over here lol
half cove
but when i run TbL.bat i get this
use this instead 
subtle token
gotta love running .bat files from ppl on the internet 
well its doing things
OH IT WORKED
omg ive been on this for like 4h now ...............
@half cove i wanna hug you right now
haha
half cove
xd
subtle token
so the model was trained at 250 epochs and i see big dip there at 6.6k, does this means this where the model would be best?
molten pecan
so the model was trained at 250 epochs and i see big dip there at 6.6k, does thi...
Yup
After the OT point it gets too fixated on the dataset and learns undesired data
Use the model closest to before the OT point
molten pecan
Also I love how you have girls kissing as your banner >_<
subtle token
ahhhh
now the question is how do i know what pth is that
i saved every 50 Epochs, did i do a big mistake? xD
molten pecan
i saved every 50 Epochs, did i do a big mistake? xD
Hmmmm just a tiny mistake
fluid matrix
subtle token
molten pecan
Zoom in on the graph and put your mouse on the OT point. Note down the step number
Use the .PTH closest to before the OT point
The checkpoints will be saved with this format:
name_epoch_step.pth
Example
arianagrande_e60_s120.pth
You got it?
subtle token
is that on the weights folder or elsewhere?
molten pecan
is that on the weights folder or elsewhere?
Yea
subtle token
thats the closest i have 
molten pecan
thats the closest i have <:l_sad:722972260369170502>
You sure?
If you set the save frequency to 50 then there should be a closer checkpoint
subtle token
yes coz i set to save every 50 Epochs...i didn't know about this... 😦
molten pecan
Is there a s5850 one?
subtle token
no
half cove
If you set the save frequency to 50 then there should be a closer checkpoint
if they save every 50 epoch it will only save every 50
molten pecan
thats all of them
Well, use the 7k one
molten pecan
if they save every 50 epoch it will only save every 50
Yeah, I messed up with step 😭
half cove
im confused why you thought this
molten pecan
lmfao
I miss helping people on help chats
I have been too busy with the RVC documentation
half cove
subtle token
damn im so sad 😦 but the voice sounds rly good using the last one, but now im kinda like...i want to know if it could sound better, almost tempted to do another retrain and save every epoch lol
molten pecan
damn im so sad 😦 but the voice sounds rly good using the last one, but now im k...
Oh probably there would be an unnoticeable difference
No worries
half cove
it shouldnt matter much
subtle token
thats how it looks, ... P.S: was my first model lol
what range is considered "good", or is there no like...hum...guideline range for what to aim
half cove
what is mel and d total lookin like
subtle token
Congrats on making your model ^-^
also thank you 😄 im really happy with the result
molten pecan
also thank you 😄 im really happy with the result
yw! Have fun
subtle token
what is mel and d total lookin like
now you are making questions that make me feel dumb
hold
thats Mel?
i dont know about G, theres a ton of "G" lol
half cove
it hasnt overtrained xd
subtle token
it hasnt overtrained xd
wait you telling me it could've trained more? D:
like more than 250 epochs?
half cove
usually you should test different save points
subtle token
are those any of what you were saying about G?
azure marshBOT
RVC Guides (How to Make AI Cover)
subtle token
oh G total was the first graph i showed
rare gobletBOT
Ayo? @subtle token level 5 !!!
subtle token
where there was that big Dip
weeeeeeee purple coolor
thats G Total
let me see the other one
This is D total on top and G total on bottom
according to D total it was still going down
when i increase the smoothness
hot lagoon
Was a good fork for infec?
subtle token
i used mainline
green ivy
Maybe there’s a way I can make the zip file shorter
subtle token
sorry to interrupt, but i just test the 2 steps closer to that g total dip and compared to the final one which is slightly higher and the final pth is actually the one that sounds better
i tested with an audio, no on w-okada obviously, so this just got more confusing to me 😂
oh well ill get the hang of it
subtle token
i mean the voice sounds amazing, i literally mimicked the real person talking and me talking and i can barely tell the difference
green ivy
I’m trying to figure out how to make the zip file smaller.
half cove
I’m trying to figure out how to make the zip file smaller.
upload it to google drive and let me try
subtle token
only problem is, i CAN'T laugh or do excited cadances, because it breaks the voice on okada entirely
dunno if thats an okada limitation or just a training limitation in general
half cove
models are usually bad with things like laughs
green ivy
upload it to google drive and let me try
You can upload it hugging face?
half cove
You can upload it hugging face?
ill try
green ivy
Okay
half cove
models are usually bad with things like laughs
you can try laughing like your saying it
subtle token
you can try laughing like your saying it
what do you mean?
coz ive been trying to "fake laugh" to get a good laugh output on okada and i can't figure anything out xD
half cove
what do you mean?
hard to explain
subtle token
Also if i do a loud and excited noise it simply cuts off the audio
half cove
like haha but say it
subtle token
the whole point was to try to do a video for friends, but i wanted to include laughter in there and excited sounds .-.
let me see if i can understand what you mean
OH
I FOUND A WAY
laugh with your nose
this is impossible to explain
but it works
D:
half cove
i guess like breathing kinda?
subtle token
also maybe this should be better to ask on #🔍│help-w-okada , but if i dont have Echo and Sup2 on theres always feedback 😦
green ivy
I just click on"New Model", somehow it kept asking me to download it.
green ivy
works for me https://huggingface.co/poiqazwsx/test2131/blob/main/Sunset_Shimmer....
Wait, what's your username on Huggingface?
half cove
poiqazwsx
half cove
Go on the website on pinned messages and make sure you did it exactly the same as the part that's says (how to upload to huggingface)
green ivy
I did it rb
I think we have to find a way to make the zip file smaller. Btw, I don't see it though.
rare gobletBOT
Ayo? @green ivy level 3 !!!
green ivy
I did it rb
Wait, what will happen if I ONLY upload the index file on hugging face?
lethal light
is there any way to make the conversions less autotuned?
half cove
Wait, what will happen if I ONLY upload the index file on hugging face?
The formatting will be incorrect so if you try downloading it from a link downloader for example ila rvc it won't work and to apply for model maker you also need to format it correctly
green ivy
The formatting will be incorrect so if you try downloading it from a link downlo...
Can you explain that of how do I do that?
molten pecan
Poopmaster go to bed 😭
half cove
Then find the bit about uploading to huggingface
half cove
Poopmaster go to bed 😭
Nuh uh
half cove
green ivy
Go in the pinned messages
Which ones do you mean exactly?
green ivy
Look at the website Julia posted
Is it the one message that's posted in 02/03/2024?
graceful jolt
whats the main cause of fragmented audio when creating these? I used the taylor swift file and its not coming out as clear as examples used on here? does it all come down to having a really clean reference acapella?
serene wigeon
in Applio, when it says "pitch" is that +/-12 for exactly one octave?
rare gobletBOT
Ayo? @serene wigeon level 5 !!!
fierce stirrup
mines is laggy
clear snow
do i have to have the command prompt open to use the voice changer?
forest sluice
in Applio, when it says "pitch" is that +/-12 for exactly one octave?
yeah basically
visual iris
Hello, I successfully created my first Voice Model today but now have question.
rare gobletBOT
Ayo? @visual iris level 1 !!!
pearl perch
my virtual cable is really quiet
azure marshBOT
RVC Guides (How to Make AI Cover)
lyric star
how i get des models more cleaar
rare gobletBOT
Ayo? @lyric star level 1 !!!
lyric star
?
oblique owl
Can someone explain me better about tensorboard, the guide you have is very vague.
timid remnant
When it comes to determining what .pth file to use, is the lowest value possible best, or the lowest value just prior to overtraining?
Because I see the lowest value happened a bit earlier during my training during a spike down, and further into the training, it never dropped that low again.
Also how much silence in voice clips should there be? Would it be ideal to get rid of all silence even if it's just for fractions of a second?
rare gobletBOT
Ayo? @timid remnant level 3 !!!
north torrent
So i have about 50 extremely short clips of my friend bunched up into about 5 minutes so far, with more for me to put together. the clips range from 1 to 3 seconds long, will this affect my model because of it cutting off alot and swapping?
heres what i have so far:
north torrent
will it work for making an ai or nah
weary egret
is there a way to make it so i dont hear the changer?
lost relic
is this a good graph just 50 minutes sample and 900 epoch
azure marshBOT
RVC Guides (How to Make AI Cover)
rare gobletBOT
Ayo? @half fjord level 1 !!!
jaunty shale
Question about applio
What is the difference between rmvpe and rmvpe+? (min pitch and max pitch and what it does)
proper shale
Question about applio
What is the difference between rmvpe and rmvpe+? (min pitc...
rvmpe+ just has a pitch threshold
it limits the max and minimum pitch possible
half cove
try using sup2 and messing with s thresh ig
also make sure ur using the right settings for ur gpu
barren locust
Any idea why no matter what I do I get same result when making model with Mangio? It sounds the same with 2/60 min input audio, or with 20/250 epochs. I followed tutorial on yt and I do everything right, but it never works and I'm pissed rn
proper shale
Wait what does it sound like? Send a result here
barren locust
I can't send audio file there
proper shale
Ya need to be level one
barren locust
Can I dm you?
rare gobletBOT
Ayo? @barren locust level 1 !!!
barren locust
Oh nvm
proper shale
Now you can post
barren locust
That's what it sounds like
This is original audio
And I trained AI on this
And it sounds the same no matter what I use to train it with
barren locust
Thanks, I'll try it
proper shale
Let me know if it solves it :)
lilac parrot
Let me know if it solves it :)
yo can u help me setting up a ai girl voice
proper shale
yo can u help me setting up a ai girl voice
just read the voice changer guide
lilac parrot
alr appercite it
barren locust
Let me know if it solves it :)
Sure!
proper shale
alr appercite it
yw :)
oblique owl
How was the quality? Do you believe it can be considered HQ?
proper shale
How was the quality? Do you believe it can be considered HQ?
considering SanInPlay's mic, yeah this sounds good
no bg noise, no artifacting
good job ig
oblique owl
Tks mod
proper shale
Yw :)
@dense sapphire
try this process, but with ur mainline folder
and write go-web.bat
dense sapphire
it doesnt work
C:\Users\user232\Desktop\RVC1006Nvidia>pause
it says to Press any key to continue . . . and doesnt open anything
subtle token
is this a good graph just 50 minutes sample and 900 epoch
I am no expert and since nobody answered , but from what i know thus far judging from the graph it started overtraining at 90k step and again even worse at around 170k steps. but generally speaking , it looks very low and it should be a very good model even at the end, probably can't tell the difference on those, you would have to test it
lost relic
I am no expert and since nobody answered , but from what i know thus far judging...
It's quite good there's no robot sound there Thank you
subtle token
It's quite good there's no robot sound there Thank you
yeah anything on the range of 35 its pretty much "perfect", can barely notice a diference when it starts flatlining around that, and u managed to go lower ^^
proper shale
C:\Users\user232\Desktop\RVC1006Nvidia>pause
u should write go-web.bat tho
hm
im being dumb
dense sapphire
"struct.error: unpack requires a buffer of 44 bytes
C:\Users\user1111\Desktop\RVC1006Nvidia>pause"
rare gobletBOT
Ayo? @dense sapphire level 2 !!!
proper shale
honestly i got no clue
hexed shard
hey, is there a good method to export only one speaker from a dialogue audio? (like podcast of 2 speakers)
(that is easier than do it myself in the normal way)
sand iris
"struct.error: unpack requires a buffer of 44 bytes
C:\Users\user1111\Desktop\R...
is there more context to the error ?
proper shale
(that is easier than do it myself in the normal way)
yeah uh sadly no, the only one that worked broke down
sand iris
give full error msg pls
hexed shard
yeah uh sadly no, the only one that worked broke down
what was it?
proper shale
what was it?
was a colab, it did the diorization part that you want to do basically
but... broken
since like last year
hexed shard
do u remember its name?
rare gobletBOT
Ayo? @hexed shard level 1 !!!
barren locust
Let me know if it solves it :)
It works now! Thanks!
proper shale
It works now! Thanks!
You're welcome :)
proper shale
do u remember its name?
It was called "Ultimate AI Dataset Collector"
barren locust
You're welcome :)
How many epochs should I do with 55 min audio?
proper shale
How many epochs should I do with 55 min audio?
go for 500, every dataset length
this can be changed in case it doesnt overtrain ofc
hexed shard
It was called "Ultimate AI Dataset Collector"
sounds amazing 😦
brittle wing
I tried to open the EasyGUI but it doesn't work
It says this:
Errno 2] No such file or directory: '/content/RVC'
/content
python3: can't open file '/content/infer-web.py': [Errno 2] No such file or directory
proper shale
I tried to open the EasyGUI but it doesn't work
yeah the UI is restricted for google colab pro users
and even then, id just use ilaria rvc if you want to convert vocals or rvc disconnected for training
ilaria rvc
https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing
rvc disconnected
https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
https://aihubdocs.github.io/en guides for everything
proper shale
yw :)
green ivy
Morning guys, I’m still stuck of exporting the Sunset Shimmer voice model. I definitely need some help!
😫☹️
proper shale
give us feature extraction tho
winged grotto
this?
proper shale
yeah now you'd need to reexport ur dataset, folder renamed to "ana"
then zip and do everything again, set experiment_name to ana too
proper shale
Another question,i have gtx 40 series, is it compatible with local traning?
yeah definitely, what 40 series one tho?
4060?
winged grotto
yeah definitely, what 40 series one tho?
4090 🤓
winged grotto
okay better i use local train right?
proper shale
heres your tutorial for mainline, install that and you're good
proper shale
I thought it was uncompatible
😭
winged grotto
yea i followed that tutorial for colab
proper shale
the only "incompatible" cards are
10 series cards
and even those can train, its just slow
also
amd and intel cards
proper shale
yw :)
azure marshBOT
RVC Guides (How to Make AI Cover)
gusty kestrel
proper shale
guys please use #🔍│find-models
winged grotto
If i stop training while it still has more epocs to do, will model be generated?
proper shale
If i stop training while it still has more epocs to do, will model be generated?
yeah
how many epochs r u on
winged grotto
These are the checkpoints?
proper shale
yes u can stop
proper shale
These are the checkpoints?
yeah those r the checkpoints
winged grotto
So I just close tab?
proper shale
are ya sure the d g files have already saved?
if not, then wait
if yes, yeah u can stop training
proper shale
And whats the difference between d and g files?
d is for discriminator, g is for generator
both are essential to continue the model training
and to train the model itself
the usable models should be in the weights folder
winged grotto
I have no such folder
proper shale
unless you disabled the "Save small finished model to weights"
winged grotto
It might be inside assets
yes, i have that folder 😂
I didnt know the path
But they are all old archives
I may have hit the settings you said
proper shale
😭
winged grotto
Welp, start over again 🤣
rare gobletBOT
Ayo? @winged grotto level 4 !!!
proper shale
And how many epochs are necessary?
It depends lol
For me I just set the total epochs to 500 and go to that, if it overtrains I'll just go back to the non overtrained version
If it doesn't overtrain I continue to 750, then 1000, etc
proper shale
Yw :)
proper shale
Ig just move the assets folder over to the new install
brittle wing
I’ll try it thanks
knotty parrot
anyone struggling with the NoWebUI Colab?
proper shale
Huggingface is currently down, that might be why
odd galleon
hi, [Voice Changer] Web Server Launch Exception, Expecting value: line 1 column 1 (char 0) and the app doesn't load, tried restarting, any ideas?
knotty parrot
Huggingface is currently down, that might be why
thanks for the heads up lol!
ionic canopy
I get this error when trying to use the inference feature on my audio
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\routes.py", line 437, in run_predict
output = await app.get_blocks().process_api(
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1349, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1283, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 2586, in postprocess
file_path = self.audio_to_temp_file(
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 360, in audio_to_temp_file
temp_dir = Path(dir) / self.hash_bytes(data.tobytes())
AttributeError: 'NoneType' object has no attribute 'tobytes'
odd galleon
hi, `[Voice Changer] Web Server Launch Exception, Expecting value: line 1 column...
okay nvm, tried deleting the strored_settings.json but it seems like the whole app is broken, will reinstall
sand iris
okay nvm, tried deleting the strored_settings.json but it seems like the whole a...
not needed
huggingface is down, app is really weird
and apparently requires to contact huggingface servers no matter what...
if huggingface goes back online, it'll work again
knotty parrot
if huggingface goes back online, it'll work again
i was confused for a bit till someone said something haha
rare gobletBOT
Ayo? @knotty parrot level 2 !!!
odd galleon
I think i will reinstall anyway, i've installed the wrong version since I have rtx 3070
ionic canopy
if huggingface goes back online, it'll work again
Hey, do you think you could help me out?
sand iris
with ?
ionic canopy
I keep getting an error running rvc locally. I am trying to use inference for one of my models
But I get this error
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\routes.py", line 437, in run_predict
output = await app.get_blocks().process_api(
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1349, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1283, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 2586, in postprocess
file_path = self.audio_to_temp_file(
File "C:\Users\james\OneDrive\Desktop\bg3-ai-voicetoolkit-main\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 360, in audio_to_temp_file
temp_dir = Path(dir) / self.hash_bytes(data.tobytes())
AttributeError: 'NoneType' object has no attribute 'tobytes'
I have encountered it in the past, but I got no clue what it means
winged grotto
Are index files important when using voice models?
proper shale
Are index files important when using voice models?
They kinda are, as it applies the accent of the character on the result
proper shale
should be in logs... hmmm
proper shale
maqybe i forgot to hit the index button
...oh
wide saddle
yoo im getting a problem with the voice changer
download sample catalog. samples_0003_o2.json
idk what that means
can anyone help me?
sand iris
download sample catalog. samples_0003_o2.json
hugging face is down again lol
wide saddle
hugging face is down again lol
bruh you serious rn
OMG BRUH
fired
brooo
NAAAA
FIRED BUY THE HUGGING FACE COMANY BRO
sand iris
it needs to download files on first launch and contact huggingface on subsequent launches as well because it's ass
sand iris
surprised it went down in the first place, but damn it must be a challenge to have that back online
wide saddle
so we just got to wait on them again?
rare gobletBOT
Ayo? @wide saddle level 5 !!!
sand iris
ya...
civic onyx
hot lagoon
in is there a way to stop training cuz I dont see the stop training thing on it?
willow turtle
Is there any downsides to using replay to make cover?
severe isle
It doesnt let me use Google Drive Models, can somebody help?
rare gobletBOT
Ayo? @severe isle level 1 !!!
severe isle
im using this https://huggingface.co/spaces/Clebersla/RVC_V2_Huggingface_Version
But when i dowload a model, it doesnt load, help??
Using Colab says this, but i dont know how to fix it!
half cove
im using this https://huggingface.co/spaces/Clebersla/RVC_V2_Huggingface_Version...
try ila rvc
-colab
half cove
the colab version is faster but huggingface works
the problem is google messed with google downloads
severe isle
Thanks
half cove
so old forks wont be able to download anymore
severe isle
Ilaria RVC doesnt let me use Drive Links
"ERROR - The download failed. Check if the link is valid."
Applio seems to work!
azure marshBOT
RVC Guides (How to Make AI Cover)
north torrent
So i have about 50 extremely short clips of my friend bunched up into about 5 mi...
any idea for this?
rare gobletBOT
Ayo? @north torrent level 3 !!!
proper shale
any idea for this?
try and see
north torrent
try and see
how many epochs do u think
brave garnetBOT
⠀
Settings for Nvidia GPUs
F0 Det.: rmvpe (suggested for all series)
RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra
Advanced Settings
Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low
⠀
Made with love by Ilaria 💖
gleaming viper
this app does not work for CS2 anymore sadly
spring orchid
anyone having issues with colab rvc disconnected ?
unborn berry
anyone having issues with colab rvc disconnected ?
server is having an aneurysm
it'll be fixed eventually
brittle wing
spring orchid
ohh thanks
vast cliff
RVC 2 Servers down?
tired vessel
This is what I'm getting :X
brittle wing
its back now.
vast cliff
poggers
brittle wing
Is there an easier and faster way to do AI convers? I'm using HuggingFace to do the convers, but it's taking about 400 seconds.
tame grail
is there a website where you can use the voice ai's from this discord for text to speech?
dawn nebula
is there a website where you can use the voice ai's from this discord for text t...
elevenlabs
dawn nebula
Is there an easier and faster way to do AI convers? I'm using HuggingFace to do ...
use colab
brittle wing
My colab is taking much longer.
brittle wing
Before he was faster.
brittle wing
https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sh...
This is new...
How to use?
dawn nebula
just run the cells
brittle wing
I did this, added the model and audio, it's almost 200 seconds... 🤡
rare gobletBOT
Ayo? @brittle wing level 2 !!!
tame grail
@dawn nebula do you know where in eleven labs i'd have to go tot upload one of the voice ai's from the library on this discord?
brittle wing
Does the audio need to be .wav? To be faster.
molten pecan
Awww wait who sent this?
Que dices? Agregamos una guía de piratería?
😭😭
Uh no mandé la captura completa
I know we can't be able to add all that in the aihub docs site, can we get a rentry link maybe?
love what you guys are doing! ❤️
exotic mural
i got two files from training RVC
one called D_3800.pth and one called G_3800.pth
which one do i use? i tried both of them in w-okada client but the client sees my model as blank
rare gobletBOT
Ayo? @exotic mural level 1 !!!
molten pecan
i got two files from training RVC
one called D_3800.pth and one called G_3800.pt...
Those aren't usable models
exotic mural
is it the one in assets/weights?
molten pecan
is it the one in assets/weights?
Yeah
molten pecan
alright tysm !
yw =3 for more help you can check our website https://aihubdocs.github.io/en/ =3
azure marshBOT
RVC Guides (How to Make AI Cover)
tall ocean
How is it possible that I trained a model that had an hour long dataset (grey) and the current model I am training (blue) is already generalizing better with only 10 minutes of data?
dawn nebula
bro rvc sucks with long datasets
oak echo
how do i start making my own voice models and how to upload them to make them into the files used to upload them into the client
finite galleon
How is it possible that I trained a model that had an hour long dataset (grey) a...
My shit hit 30 at like 600 steps and it was only 2 minutes
silver bobcat
how do i fix error no CUDA GPUS are available
sand iris
how do i fix error no CUDA GPUS are available
context pls
are you running locally or on colab ?
If it's locally, do you know if your pc has a GPU ?
silver bobcat
im on colab
sand iris
you sure you picked the "t4 gpu" runtime option ?
silver bobcat
i think my friend didnt,thanks
brittle wing
can someone help me, i downloader the files for a model but i don't know how to download the model on hugging dace
wtf did i write lol btw i need help
azure marshBOT
RVC Guides (How to Make AI Cover)
proper shale
how many epochs do u think
go for 500, every dataset length
if it doesn't overtrain go for 750 then 1000 and so on
grave brook
is there any notebooks i could use for batch infer? cuz my laptop from 2016 isnt powerful enough to
i saw that the RVC v2 Disconnected's repo that the notebook uses can use CIL to infer though im not sure if it allows batch conversion via sepcifying audio folder
wanton void
I have a question i use the voicechanger but in the setting of gpu i only see cpu but i have gpu someone can help?
proper shale
is there any notebooks i could use for batch infer? cuz my laptop from 2016 isnt...
good question ngl, i think the best one is that one kaggle notebook with multiple forks n stuff... not even sure how you'd use that ngl, but I do know mainline and mangio have batch infer options
@glad zealot can ya send that link please?
also if its not a lot of files u could just use ilaria rvc ig
proper shale
I have a question i use the voicechanger but in the setting of gpu i only see cp...
what GPU do you have
wanton void
AMD Radeon RX 6600 XT
wanton void
so i cant use gpu on the voicechanger
rare gobletBOT
Ayo? @wanton void level 1 !!!
wanton void
this is the right one
thanks i will check mate
proper shale
also, you need to convert downloaded models to onnx, check the guide:
https://rentry.co/VoiceChangerGuide#uploading-models-amdintel-gpus-and-fixing-its-laggy-issue
wanton void
also, you need to convert downloaded models to onnx, check the guide:
https://r...
Thanks bro
proper shale
Thanks bro
Yw :)
Let me know if it works
wanton void
Yw :)
Let me know if it works
K
I see the option cpu gpu 0 gpu 1 gpu 2 and gpu 3
what i need to choose?
@proper shale
proper shale
I see the option cpu gpu 0 gpu 1 gpu 2 and gpu 3
normally it's gpu0
wanton void
so gpu 0 its my gpu?
molten pecan
normally it's gpu0
MJ!!!!!!
https://aihubdocs.github.io/en/ the new update came out!!
Check it out!
And also
https://aihubdocs.github.io/es/ Spanish version is out ;3
wanton void
youre talk to me?
proper shale
so gpu 0 its my gpu?
probably yes
proper shale
youre talk to me?
nah she was talking to me
oblique owl
How was the quality? Do you believe it can be considered HQ?
wanton void
and in the F0 det what i need to choose
proper shale
How was the quality? Do you believe it can be considered HQ?
yeah this sounds filtered and has a lil artifacting
proper shale
and in the F0 det what i need to choose
rvmpe_onnx is best
proper shale
yeah this sounds filtered and has a lil artifacting
so no, I don't believe this can be considered HQ
oblique owl
yeah this sounds filtered and has a lil artifacting
So it's not HQ?
proper shale
yeah no
oblique owl
but is it due to lack of epoch or is it because of the dataset?
proper shale
but is it due to lack of epoch or is it because of the dataset?
dataset
wanton void
i can download model from weights.gg?
proper shale
i can download model from weights.gg?
yeah
rare gobletBOT
Ayo? @wanton void level 2 !!!
wanton void
thanks
brittle wing
and what chunk and extra is best? mate
ey can u send me the donwload file RVC
wanton void
ey can u send me the donwload file RVC
ye but what is your gpu
brittle wing
how i see that
proper shale
how do i install rvc
what rvc
brittle wing
amd ryzen 9
proper shale
and what chunk and extra is best? mate
keep extra at 8192 ig, chunks you gotta experiment, anywhere from 100 chunks to 200 might be good
brittle wing
6900hx
proper shale
amd ryzen 9
bruh thats the cpu
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
pls
wanton void
i think he wants like me
proper shale
what's your gpu
brittle wing
amd ryzen 9
brittle wing
how do i set it up?
proper shale
...the guide explains it
sand iris
jesus
brittle wing
i still dont understand
proper shale
again, take your time
brittle wing
idk
proper shale
no need to rush anything
rain bronze
the training g/total is practically flat now
rare gobletBOT
Ayo? @rain bronze level 1 !!!
rain bronze
but the actual value keeps going up and down
i set the smoothing to 0.987 as said in the guide
it's been going for 4k steps now
proven acorn
is it actually possible to make models on mobile
rain bronze
it's been going for 4k steps now
my question is that is this sign of overtraining
proper shale
yeah no, that might be too early to say its overtraining yet
unless you're like
in ov2
proper shale
just wait and see if it starts increasing a lot
proper shale
the pretrain type you mean right
yea
rain bronze
just wait and see if it starts increasing a lot
how long before i can be sure of overtraining
it's slowly going downhill now
proper shale
how long before i can be sure of overtraining
it really varies
but once you see that it's gone up a lot and doesn't seem to come back down, it most likely has been overtraining
but you can try those models tho
also, check mel and kl
brittle wing
i did it how i can use the voice models ?
proper shale
should i use the resampling option on rvc
yeah no
proper shale
i did it how i can use the voice models ?
just upload them and use them
check the guide, please
uneven mirage
I can't hear myself at all
i'm not sure what's causing it, i have my mic as my input, and as my output i have VB virtual cable
brittle wing
how do i uploud
brittle wing
just upload them and use them
check the guide, please
ye but the go bath thing dont satrt
start
it says it paused please press any key to close
molten pecan
ye but the go bath thing dont satrt
which RVC are you using?
brittle wing
hello brothers, I found a juice wrld ai on #1175430844685484042
what I need to do next
scenic gull
-rvc
scenic gull
does llaria rvc still work (the huggingface version)
molten pecan
how i asee htat
send a screenshot of your folder
rare gobletBOT
Ayo? @brittle wing level 2 !!!
brittle wing
my bad
brittle wing
here
molten pecan
how make screenshot
did you unzip it first?
brittle wing
it didnt tell me
molten pecan
try executing from the unzipped folder
brittle wing
jsut winrar
can u jsut help me in dms pls
im loslt
can i show u my screen and u help
pls
molten pecan
ok then can u go voice channel can i share my screen there
no need to do that, it's pretty easy
unzip the mangio folder first
molten pecan
then unzip it
brittle wing
i do but it show me couple of errors while unziping
molten pecan
i do but it show me couple of errors while unziping
can you send a scs?
