but imo, there's no point unless you actually need that much of power
#✨│ai-help
wispy lodge
In modded version, it will use fp16 only when possible. It's disabled for GPUs that are known to be not capable/not performant with it
stone gust
Hi pls give me the Google Collab link pls
wispy lodge
And force fp32 option is mostly in case if somehow gpu didn't get into filter or there are actual problems
analog obsidian
In modded version, it will use fp16 only when possible. It's disabled for GPUs t...
Ow i really thought it would make a quality difference xD thank goodness i asked this
stone gust
Hello! I have a question
glacial pollen
Give me the link now!!
Chill and behave.
You have a functioning brain and a search bar
stone gust
Sorry.
glacial pollen
rare gobletBOT
Ayo? @stone gust level 1 !!! 
glacial pollen
time to learn how to use discord's ui
Some people need to learn respect and when to wait if an active case's being solved
tame mica
-colab
AI HUB Docs
analog obsidian
Remember colab has a 2 hour limit
I believe u can increase this by paying tho not sure
stone gust
Which is the link there are so many
analog obsidian
Which is the link there are so many
The one i sent u is the correct one
stone gust
Okay
frigid stone
How do you figure out the sample rate of a .pth file so you know what other .pth files you can merge it with?
glacial pollen
How do you figure out the sample rate of a .pth file so you know what other .pth...
Do an infer on a clean 48khz audio, load up the inferenced audio ( model's output ) in any program that let's you see a spectrogram
and the rest should be simple. Look at the frequency response range
Or do an infer and send me the output here
frigid stone
Got it, thanks!
glacial pollen
Np
In case you can't find anything good, here's a clean 44.1 ( might not be ideal but can still work )
SynthV's AI output, if anyone asked.
( Yes, it is a good performance tester for female models - which I use quite often )
analog obsidian
How do you figure out the sample rate of a .pth file so you know what other .pth...
click edit on wokada and look for your model
the number after f0 is your sample rate
glacial pollen
They might not use w-okada tho
analog obsidian
it should say either 32000, 40000 or 48000
glacial pollen
They wanna merge the pth afteral ( ckpt fusion ) that's in rvc
frigid stone
Yeah I use RVC, but I can pick up w-okada if it lets me read the sample rate real quick
rare gobletBOT
Ayo? @frigid stone level 1 !!!
analog obsidian
They wanna merge the pth afteral ( ckpt fusion ) that's in rvc
ow man i got confused again lmao i thought this was the wokada channel
frigid stone
Thanks, bot
analog obsidian
yea just inference and u can see the sample rate aswell
glacial pollen
^ easier approach than dl'ing gigs
tho, reality is
if you can hear it not muddy / muffly
it is 48
if it's somewhat fine but lacks that " harsh ish " fidelity and clarity, especially in sibilants, it's 40khz
if it's muddy or foggy, you get the idea, it's 32khz
But then, some 40khz sets or so are used in 48khz models hence yea, better send the output here and I can help with that
analog obsidian
or just use spek (this is a inference result)
glacial pollen
Doesn't let you see the upsampling's effects tho
there are some subtle granual nuances here n there, better off If I check it in rx
analog obsidian
there are some subtle granual nuances here n there, better off If I check it in ...
how good is to merge two different voices anyways? we know merging epochs is fine but what about for example
1 voice g/total 2 voice g/total
not from the same model ofc
glacial pollen
I'd say it depends
how compatible are the voices ( timbre wise )
and what kind of ranges they learned
if one sucks at high, other doesn't
crappy abomination can occur 👀
I also wouldn't advise mixing low pitch voices with high ( say, mommy voice with high pitched anime girl voice )
Butttt, ye, experiment if you want
analog obsidian
glacial pollen
ikr
frigid stone
So I checked the sample rate of the voices given by default with RVC, kikiv1 and guanguanV1, both say they're 32k but they won't merge in Applio, I get a TypeError "cannot unpack non-iterable KeyError object"
analog obsidian
So I checked the sample rate of the voices given by default with RVC, kikiv1 and...
Because applio cant run mainline models iirc
Or at least time i tried it didnt worked xD
frigid stone
Can't even merge them because of that? Is there a better tool for merging? I basically just chose Applio because it was the first program that had merging
analog obsidian
Can't even merge them because of that? Is there a better tool for merging? I bas...
Mainline
frigid stone
Mainline RVC can merge?
analog obsidian
Mainline RVC can merge?
Yep
glacial pollen
Not sure about " applio can't run mainline models " as all models are the same, regardless of the fork
pastel oak
Wasnt the wokada models ones because theyre onnx?
glacial pollen
I'd be more opting towards 32k isn't supported for merging
There's no ui option for that nor within the code
frigid stone
Yep
What tab in the web-ui do I click into for that?
I skimmed over the web-ui for RVC a little while ago but never found one which is why I had Applio
analog obsidian
Not sure about " applio can't run mainline models " as all models are the same, ...
I wonder why last time i tried loading one of my mainline models and the results sounded like sans speaking sample
e e e
glacial pollen
cause applio itself was most likely broken
glacial pollen
What tab in the web-ui do I click into for that?
I skimmed over the web-ui for ...
will be called model utils
There @frigid stone
analog obsidian
will be called model utils
Mainline can do 32k merging?
glacial pollen
nope
glacial pollen
ignore the left side tho
analog obsidian
F
glacial pollen
Buttttt, from what I know
you can just select 40k
for 32 merges
YET I cannot promise any stability
might work fine on the outside, might be a mess inside
and some inference might come out badly, or may not
analog obsidian
Im gonna try applio again with mainline models tho, if it doesnt work again then… yea
frigid stone
I have created life, thank you very much for the assistance
analog obsidian
I have created life, thank you very much for the assistance
Yay
frigid stone
Time to dump Applio because it's not working for the one thing I downloaded it for
analog obsidian
Time to dump Applio because it's not working for the one thing I downloaded it f...
Mainline is the goat
👀
frigid stone
I didn't read hard enough to find the merge, but now that I have, it definitely is
pastel oak
RVC Mainline, if youre using the prepackage RVC1006 will not work fine with merges btw in case youre using that
glacial pollen
Time to dump Applio because it's not working for the one thing I downloaded it f...
if you want a good enhanced mainline, go for my fork
frigid stone
I'm using the one literally called "Latest" because the one in the releases doesn't work with ASIO
analog obsidian
if you want a good enhanced mainline, go for my fork
Please tell me thats fixed in your fork
hollow fox
guys ı need to train voices but that ı used for training is on google colab and it has limits, ı need free and unlimited one can anyone help with that ?
pastel oak
I'm using the one literally called "Latest" because the one in the releases does...
Ye thats mine thatll work
analog obsidian
The merging part
glacial pollen
https://github.com/codename0og/codename-rvc-fork
It does work
analog obsidian
Oof nice
glacial pollen
Merging is a part of my training routine, always
32k should work fine too, just select 40k or 48k, test stuff around
analog obsidian
RVC Mainline, if youre using the prepackage RVC1006 will not work fine with merg...
I asked that because he said this
glacial pollen
but tbf, I do not absolutely see any reason to recommend mainline over my fork
it should be either that or applio
frigid stone
The Latest one I use works fine with merges as far as I can tell
rare gobletBOT
Ayo? @frigid stone level 2 !!!
frigid stone
Thanks, bot
glacial pollen
as Mine's literally mainline but enhanced ( including model, index and audio picking
frigid stone
I tested it with the 1 merge that worked from Applio
glacial pollen
( +, recently finished moving to fumiama's base - just not updated on github yet
analog obsidian
( +, recently finished moving to fumiama's base - just not updated on github yet
Im waiting for the update 
glacial pollen
Tbf, it's nothing of a game changer if you want me to be honest
rare gobletBOT
Ayo? @glacial pollen level 19 !!!
glacial pollen
it's more so now the fork will be compatible with potential big changes fumiama could introduce
analog obsidian
I tested it with the 1 merge that worked from Applio
So applio models works in mainline? Ah thanks, so it seems my applio installation just bugged last time i tried something similar lol
glacial pollen
( more or less. Hash is a nono and won't incorporate it
red kayak
<@460577350900514837>
i was away :p
glacial pollen
i was away :p
All good
frigid stone
So applio models works in mainline? Ah thanks, so it seems my applio installatio...
Yeah, I made a merge by brute forcing whatever models actually merged, and then it worked in my RVC just fine
glacial pollen
just one thing I can say for sure what breaks most if not all merged models is
feature extraction type
i.e. Mangio x rmvpe models, might not always sound fine or work fine
frigid stone
Time to go through all my models and figure out what khz they are
glacial pollen
and, that's 200% confirmed, different precision mode trained models
are not liking each others
glacial pollen
for instance, fusing bfloat16 with fp32 resulted in a tragedy
but I wouldn't worry. Majority of models you can find are fp16 anyways
fp32 are super rare
analog obsidian
Welp hes safe because everyone trains in fp16
glacial pollen
I'd even go as far as saying at least 90% or 95% of models are fp16
hollow fox
guys ı need to train voices but that ı used for training is on google colab and it has limits, ı need free and unlimited one can anyone help with that ?
rapid knot
can anyone help me? my voice changer seems to not be working. ive correctly got everything (python and etc.) and got all the inputs outputs correctly but when i press play on a diff app (yes i have virtual audio cable) it just does not change the voice
rare gobletBOT
Ayo? @rapid knot level 1 !!!
pastel oak
can anyone help me? my voice changer seems to not be working. ive correctly got ...
Screenshot your voice changer app
rapid knot
Screenshot your voice changer app
pastel oak
Is your passthru always green when you start voice changer? does it ever become red
rapid knot
Is your passthru always green when you start voice changer? does it ever become ...
its always red
rapid knot
done
rapid knot
also do F0 det: rmvpe
where can i do that?
rapid knot
oh mb
it still does not work like i dont hear a thing on the discord test one and stuff ill go check again if i have put the wrong ones
pastel oak
it still does not work like i dont hear a thing on the discord test one and stuf...
Did you select cable as your input on discord
rapid knot
Did you select cable as your input on discord
Yup and it doesnt recognize the voice
pastel oak
Yup and it doesnt recognize the voice
Uninstall vb cable beacuse its bad and download the better one
VAC
rapid knot
alr
knotty moth
its always red
red means it picks the unprocessed voice
rapid knot
red means it picks the unprocessed voice
i see
i got it all correct but its still the same, not picking up my voice and not replaying it
zealous shuttle
tidal trail
zero gpu has a limit
rapid knot
the volume also doesnt change even when i start it
zealous shuttle
zero gpu has a limit
So the A100 is for premium user only?
rare gobletBOT
Ayo? @whole rain level 1 !!!
flint geyser
yes?
flint geyser
how can i use the google voice colab
you mean rvc?
whole rain
you mean rvc?
i belive it has a drake pic
flint geyser
this is way too old sorry
flint geyser
i belive it has a drake pic
check https://aihub.wtf/docs for help
whole rain
whole rain
check https://aihub.wtf/docs for help
what should i use then for voice clone? with samples
flint geyser
what should i use then for voice clone? with samples
check the docs
do you want to clone someones voice or use a cloned voice?
whole rain
do you want to clone someones voice or use a cloned voice?
clone someones voice
flint geyser
either follow this:https://docs.aihub.wtf/essentials/how-to-make-voice-models/
or pay people to make it (e.g. me and others)
knotty moth
So the A100 is for premium user only?
there are more affordable gpu rental alternatives
tame mica
truly peak of ai hub. racism and slurs everywhere
zealous shuttle
there are more affordable gpu rental alternatives
which one?
flint geyser
truly peak of ai hub. racism and slurs everywhere
already reported to admins.
nvm
they removed timeouts for helpers
pastel oak
can anyone help the voice do not work on the voice app
!howtoask
patent trellis
cunning juniper
my voice doesnt get morph to the model, it just sound the same
i didnt download python or pytorch, is this what may be causing the issue?
rare gobletBOT
Ayo? @cunning juniper level 1 !!!
cunning juniper
i think we have the same issue with this one
violet heron
my voice doesnt get morph to the model, it just sound the same
Is passthru on?
It should be green
Send a screenshot of your settings
cunning juniper
It should be green
its red
cunning juniper
Make it green
how do you do that
violet heron
Change it to a RVC model
violet heron
how do you do that
Click it
distant thicket
came someone plssss help me
violet heron
came someone plssss help me
Don’t ask to ask, just ask
cunning juniper
Change it to a RVC model
wheres the rvc model
violet heron
wheres the rvc model
Anyone that isn’t the Beatrice one
distant thicket
Don’t ask to ask, just ask
its been so weird every time i try to use it sounds like im talking to a spirit box
violet heron
Anyone that isn’t the Beatrice one
What’s your GPU?
violet heron
its been so weird every time i try to use it sounds like im talking to a spirit ...
Adjust tune
distant thicket
Adjust tune
i talk to it and it says nothing back
orchid glade
been using the RVC - GUI and have been using models on this server but it seems like most of them sound distorted but when I hear the samples it seems fine
what could I be doing wrong 🤔
violet heron
i talk to it and it says nothing back
Did you click start?
violet heron
been using the RVC - GUI and have been using models on this server but it seems ...
RVC-GUI is outdated
distant thicket
can someone hop on a call with me?
rare gobletBOT
Ayo? @distant thicket level 1 !!!
distant thicket
Did you click start?
yeah the ui is also super laggy
orchid glade
What should I use now then ?
violet heron
can someone hop on a call with me?
Do you have the audio cable set up?
distant thicket
Do you have the audio cable set up?
i do but how do i use it?
violet heron
What should I use now then ?
-local I personally use original/mianline
azure marshBOT
-local I personally use original/mianline
Suggestions for @orchid glade
violet heron
i do but how do i use it?
Set output on Okada to the cable and input on discord to the cable
distant thicket
Set output on Okada to the cable and input on discord to the cable
i have not even tierd it on discord its not even working on its own
violet heron
i have not even tierd it on discord its not even working on its own
Send a screenshot of your settings
Also what’s your GPU
distant thicket
Also what’s your GPU
1660
orchid glade
rip
distant thicket
distant thicket
oh.
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
distant thicket
so i need the other one
violet heron
> This interaction has expired, use the command `/guides realtime` if you wish t...
Read the first one
Should have the NVidia download
violet heron
rare gobletBOT
Ayo? @distant thicket level 2 !!!
jade shore
why does my ai model not have an index!
????
I tried to make a salad fingers ai cover and it dont have an index!
rapid knot
i think we have the same issue with this one
i mean i have python and pytorch but no idea why it doesnt work
rare gobletBOT
Ayo? @rapid knot level 2 !!!
cunning juniper
i mean i have python and pytorch but no idea why it doesnt work
still not working until now?
rapid knot
still not working until now?
nope, im trying to reinstall rn
distant thicket
what do i set moniter to?
jade shore
I tried to make a salad fingers ai cover and it dont have an index!
can someone help me?!
undone vault
hey everybody, just getting into this AI voice stuff. Can somebody point me to a guide of how to use a voicemodel that i get from this server?
I have heard that there may be different methods for different models, here is the one I am attempting to use: #1175498017697181858 message
I just need the elevenlabs adam voice, but they removed it from the site. If this voice model I linked here is bad for whatever reason, feel free to point me to any others you happen to be aware of
pastel oak
can someone help me?!
You dont need an index
distant thicket
unzip the folder first
im talking and nothing is happending
cunning juniper
nope, im trying to reinstall rn
hello how do you uninstall it?
jade shore
You dont need an index
really?o
rapid knot
hello how do you uninstall it?
delete all the files?
cunning juniper
delete all the files?
just delete it?
pastel oak
Voice model goes into assets -> weights folder
Index goes into logs folder
rapid knot
just delete it?
i mean its what i did, still reinstaling tho ill keep u up with the results and if doesnt work lets both try to get help
jade shore
there isnt an index in the model folder i download
pastel oak
there isnt an index in the model folder i download
Not the end of the world
Index is accent of model , not needed
cunning juniper
i mean its what i did, still reinstaling tho ill keep u up with the results and ...
sure thanks
pastel oak
hey everybody, just getting into this AI voice stuff. Can somebody point me to a...
For making ai covers ("Inference") check this out
#✨│ai-help message
jade shore
Index is accent of model , not needed
Thanks i did it
wispy lodge
i mean i have python and pytorch but no idea why it doesnt work
You don't need to install pytorch and even python
Show screenshot with settings and command line that opens when you run start_http
cunning juniper
i mean its what i did, still reinstaling tho ill keep u up with the results and ...
could u check dm's? thanks
rare gobletBOT
Ayo? @cunning juniper level 2 !!!
wispy lodge
Set chunk to 192 and extra to 8192
undone vault
For making ai covers ("Inference") check this out
https://discord.com/channels/1...
ideally id be looking for TTS with that voice model, unless there is something I dont understand
or is the point to use one of these TTS models on the site and then convert to the desired voice via inference?
distant thicket
Set chunk to 192 and extra to 8192
how do i turn the sensitive down
distant thicket
Set chunk to 192 and extra to 8192
its picking up background noise
wispy lodge
its picking up background noise
Enable sup2 for noise suppresion
undone vault
or is the point to use one of these TTS models on the site and then convert to t...
this seems to be the answer. Result is not necessarily great, but thats probably down to the model and some fine tuning. Thanks for the help 👍
rare gobletBOT
Ayo? @undone vault level 1 !!!
sour shuttle
where can i download rvc V2?
rapid knot
Show screenshot with settings and command line that opens when you run start_htt...
sorry for the late response
wispy lodge
sorry for the late response
Which model did you load?
rapid knot
Which model did you load?
just a predownloaded one (if thats what your asking)
rapid knot
prob because i redownloaded it ill tell u if it works after or not
rapid knot
The download was not complete
Now its this and still doesnt seem to be working
rapid knot
The 3050 laptop one nvm just saw it
yup
undone vault
For making ai covers ("Inference") check this out
https://discord.com/channels/1...
any resources for making the model be a little more excited/yelling? or would that need to be a modification to the model itself?
rapid knot
Then you need to download the cuda toolkit thats listed on there
uh where exactly is the cuda toolkit?
and just now i uploaded a voice model but it seems to give this when i try to switch to it
wispy lodge
The problem might be that you already have cuda toolkit installed, but it's incompatible
rapid knot
The problem might be that you already have cuda toolkit installed, but it's inco...
i see, so can i just not use new models?
wispy lodge
i see, so can i just not use new models?
You can try editing start_http.bat and add the following line in the beginning
set CUDA_PATH=
rapid knot
You can try editing start_http.bat and add the following line in the beginning
`...
where's start_http.bat? I cant seem to find it on the MMVC something folder
rare gobletBOT
Ayo? @rapid knot level 3 !!!
wispy lodge
How did you start the voice changer?
rapid knot
How did you start the voice changer?
by start_http ?
wispy lodge
Well, that's the file, yeah
wispy lodge
Right-click it > Edit
rapid knot
Right-click it > Edit
alr did that ty
rapid knot
Might have to do "Open with" and select notepad
yup it was on notepad
well how do i get the voicechanger to work now?
OH NVM IT DOES
sick coral
Help pls
C:\Program Files (x86)\MMVCServerSIO>MMVCServerSIO.exe -p 18888 --https false --content_vec_500 pretrain/checkpoint_best_legacy_500.pt --content_vec_500_onnx pretrain/content_vec_500.onnx --content_vec_500_onnx_on true --hubert_base pretrain/hubert_base.pt --hubert_base_jp pretrain/rinna_hubert_base_jp.pt --hubert_soft pretrain/hubert/hubert-soft-0d54a1f4.pt --nsf_hifigan pretrain/nsf_hifigan/model --crepe_onnx_full pretrain/crepe_onnx_full.onnx --crepe_onnx_tiny pretrain/crepe_onnx_tiny.onnx --rmvpe pretrain/rmvpe.pt --model_dir model_dir --samples samples.json
Traceback (most recent call last):
File "MMVCServerSIO.py", line 33, in <module>
File "mods\log_control.py", line 83, in initialize
File "logging_init_.py", line 1169, in init
File "logging_init_.py", line 1201, in _open
PermissionError: [Errno 13] Permission denied: 'C:\Program Files (x86)\MMVCServerSIO\vcclient.log'
[10304] Failed to execute script 'MMVCServerSIO' due to unhandled exception!
fierce isle
hello, i have a question and a little help about the rvc how will i run it again after closing it?
is it the same step when i opened it on the file i downloaded?
distant turtle
hello, i have a question and a little help about the rvc how will i run it again...
what are you using
why this error?
distant turtle
why this error?
all colab versions give me problems, errors, why?, locally the problems are different, but colab shouldn't cause problems, I've always used it
pastel oak
why this error?
Problem seems to be voice model
What did you put in for voice model
What path or link
pastel oak
hello, i have a question and a little help about the rvc how will i run it again...
Thats an odd obvious question or am I misunderstanding
You open the file the way you started it the first time yes
fierce isle
Thats an odd obvious question or am I misunderstanding
You open the file the way...
You answered my question, thank you
distant turtle
Problem seems to be voice model
I cloned the voice as usual and put it on huggingface
pastel oak
Help pls
C:\Program Files (x86)\MMVCServerSIO>MMVCServerSIO.exe -p 18888 ...
Lots of things that could be the case
- Disable antivirus program
- If you did indeed have an antivirus program running, then delete the MMVCServerSIO folder again (you DONT have to delete the zip file with the bunch of numbers behind it)
And then extract/unzip it again with the antivirus disabled - What gpu do you have
pastel oak
I cloned the voice as usual and put it on huggingface
Whats the URL path you put in
severe bay
can you download minecraft parkour anywhere for free?
pastel oak
What does that have to do with AI
distant turtle
Whats the URL path you put in
This seems strange to me since I always do this, thanks for the reply
pastel oak
This seems strange to me since I always do this, thanks for the reply
remove the ?download=true from the url when you insert it
distant turtle
remove the ?download=true from the url when you insert it
ok I'll try, I don't use colab much, I used local more, but it gives me a strange error
on mangio rvc, locally it gives me this error, I use an AMD
pastel oak
on mangio rvc, locally it gives me this error, I use an AMD
Id recommend you get mainline instead
there you run go-web-dml.bat
I unfortunately dont recognize the error on your mangio local either
distant turtle
<https://huggingface.co/Shadicti/RVCLatest/blob/main/RVCAMD_IntelLatest.zip>
ok I'll try, thanks
turbid flax
for creating AI models, would you include recordings from different microphones in the dataset or not? would it make rvc perform better or worse?
since people sound different with different microphones
with perform i mean would it sound more or less realistic
pastel oak
for creating AI models, would you include recordings from different microphones ...
For realtime purpose, you should definitely only have ONE microphone used
For RVC, at least for me, it has not made much of a difference assuming the person doesnt suddenly sound completely different. Some difference is okay
turbid flax
thanks!
pastel oak
with RVC i mean inferencing/ai covers
turbid flax
yeah i know
glacial pollen
Posting it here in case someone uses 1.0.0 of my rvc rtvc as I've pushed the update:
https://github.com/codename0og/rvc-realtime-voice-changer/releases/tag/v1.1.0
And here's instructions on what's what etc for new users.:
https://github.com/codename0og/rvc-realtime-voice-changer
plain fulcrum
Guys which one is more important, dataset or epochs? If I wanna make my model better, should i add more audio data or increase epoch number?
glacial pollen
Both
Epochs and so, training duration
batch size
dataset
Pretty much everything is equally as important, there's no "priority"
plain fulcrum
I'm using replay ai and i don't think it has those options like batch size
I didn't see em anyway
analog obsidian
Posting it here in case someone uses 1.0.0 of my rvc rtvc as I've pushed the upd...
glacial pollen
I'm using replay ai and i don't think it has those options like batch size
Then that's on you for using such things
plain fulcrum
What should I use then? I don't have a gpu so that's a bit of an issue
glacial pollen
Well, there are other alternatives
glacial pollen
-colab
glacial pollen
-docs
glacial pollen
Have a read
There's some zero spaces or such, maybe ilaria?
Not sure 100% as I don't track these
plain fulcrum
Alright I appreciate it
glacial pollen
but if you search through this discord
you'll find lots of alternatives / solutions
plain fulcrum
And by the way do they have the ability to pause the training?
plain fulcrum
I wanna train models for a long time but I don't wanna have my laptop running for a week straight💀
glacial pollen
and appropriate files have to be kept
hold on.. Imma check if I made a guide for it in the past
plain fulcrum
Alright
glacial pollen
oh, seems like not.. welp.
uhhhh
tl;dr
- you can stop the jupyter cell or whatever you use that keeps the rvc running as long you don't see " saving optimizer state at xxxx "
n such
or any info displaying savings of generator ( G ) or discriminator ( D )
in a console or such
other than that, you need:
- your model's folder
( ur model's folder from log folder )
it has to contain everything
- most recent .pth for your model from training
There
plain fulcrum
Man I really gatta step up colab knowledge I didn't understand a single word you said😭
glacial pollen
Whenever you train, after each trained epoch there's saving taking place. Easiest way you can spot it is by noticing " saving optimizer state at epoch xxxx " or something along the line
vague abyss
What's the difference between "E3" and "Ov2"? What specifically do they mean? Sorry if asked a lot, couldn't find anything about it
glacial pollen
you can't pause during that moment because the generator or discriminator ( which have the optimizer states saved at the time ) will break
Next thing would be, each trained or actively trained model has it's folder
analog obsidian
Whenever you train, after each trained epoch there's saving taking place. Easies...
completely unrelated to the convo;
does installing the requeriments for the voicechanger messes up things in your fork? im a bit paranoid fsfs
glacial pollen
models' folders are located in " log " folder
plain fulcrum
Whenever you train, after each trained epoch there's saving taking place. Easies...
Ohh so it automatically saves itself after every epoch?
If that's the case that's nice
glacial pollen
each epoch = a produced model
plain fulcrum
Right
glacial pollen
as long you have saving frequency set to 1
glacial pollen
models' folders are located in " log " folder
now, continuing
Whenever you wanna resume the training, your model's folder has to be present in " log " folder
- the most recent epoch / .pth model / weight mode ( most recent one you had when you stopped the training )
those are located in: rvc folder / assets / weights folder / here
And visually it'd be:
plain fulcrum
Ah got it so I can continue the training from the last saved epoch
I'm barely keepin up with this
plain fulcrum
Right
glacial pollen
this is my model's folder located in logs folder
as you can see from the address bar
such have to be present
( if you like, switch colabs and so on or whatever )
and aside that, most recent .pth model
imagine this is my model's 192th epoch
so MyModel_e192_Sxxxx.pth
more or less
plain fulcrum
Wait this is a bit unrelated but I saw somewhere tht colab stopped giving access to rvc for free? I thought there was like a time limit for the training
glacial pollen
you'd have to have that .pth in:
rvc folder / assets / weights / here
and model's folder in:
rvc folder / logs / here
then when you resume the training, you only input the model's name again ( as in, " experiment name " or however the given rvc is localized ) and hit train model
all the mid-steps such as feature extraction etc. are only done once
violet heron
Wait this is a bit unrelated but I saw somewhere tht colab stopped giving access...
For webUIs yes
But there are bypasses
glacial pollen
Wait this is a bit unrelated but I saw somewhere tht colab stopped giving access...
You gotta ask others really as I stopped using colab very very long time ago
violet heron
But there are bypasses
Some of the colabs don’t use webuis or have a bypass
glacial pollen
completely unrelated to the convo;
does installing the requeriments for the voic...
Nope
it's cross-compatible
in fact
you should be fine with just my requirements for rvc
lemme confirm it tho
mild dock
hey guys i have questions, im very new to ai modeling how do you do this?
im a complete noob
violet heron
hey guys i have questions, im very new to ai modeling how do you do this?
What are you trying to do?
Make a cover or model
plain fulcrum
im a complete noob
Finally someone on my level of understanding
mild dock
A cover
plain fulcrum
Finally someone on my level of understanding
Oh no wait sry didn't mean to insult u
mild dock
What is the difference between them?
glacial pollen
lemme confirm it tho
Yup, you'll be fine
analog obsidian
Nope
omg it also fixed the weird flexasio bug i had
flexasio decided to stop working in wokada and mainline's realtime, but works perfectly fine in your fork, thank uuu 😭
mild dock
sorry im completley lost
glacial pollen
voice changer's list is actually a part of rvc's list
violet heron
What is the difference between them?
The local one will be a program installed onto your computer and it depends on your GPU
analog obsidian
Yup, you'll be fine
yay
plain fulcrum
sorry im completley lost
U can either do it online on a website or a program on ur pc
glacial pollen
So ye, as long you have all that was needed for prev version, you're fine
no need to install extras
violet heron
The local one will be a program installed onto your computer and it depends on y...
Online/cloud are via websites
mild dock
Ohhh, whichever one is easier
violet heron
Online/cloud are via websites
But they usually have some restriction of some sort
mild dock
maybe website?
violet heron
Ohhh, whichever one is easier
-spaces Ilaria RVC
azure marshBOT
-spaces Ilaria RVC
Suggestions for @mild dock
analog obsidian
yup i had everything installed before, thank goodness
violet heron
Try using Ilaria RVC it’s easy to use and fast
To put your model in go to the upload/download section and put in the models pth and index
Or the models download link
Then you upload the audio (clean vocals no music) you want to convert, select your model and convert!
glacial pollen
yup i had everything installed before, thank goodness
Awesome, have fun man ~
mild dock
Okay cool i have a model downloaded right now, do i need to convert it to anything? or can i just drag it into the upload sections right away
violet heron
Okay cool i have a model downloaded right now, do i need to convert it to anythi...
Drag into the upload section
Make sure you unzipped it since in there are the .pth and .index files
rare gobletBOT
Ayo? @mild dock level 1 !!!
mild dock
do they usually take a while to upload?
violet heron
do they usually take a while to upload?
Depends but shouldn’t take long
Might depend on your internet
mild dock
ohh okay yeah because i just dragged both in like 2 min ago and the pth one just got uplaoded
waiting for the index to competley uplaod both
okay i just uploaded the model
whats left to do?
violet heron
whats left to do?
Go to infernence
Click refresh and click your model
Then upload your clean, without music, audio and click convert
mild dock
oh cool is that it?
violet heron
oh cool is that it?
Yeah it should take less then a minute
If you want to adjust accent in can under index ratio you can in advanced settings
mild dock
ohh okay cool! thank you guys i apreciate it
crude bolt
guess im stuck ... for some unknown reason i can not train anymore with applio - it worked before (4 hrs ago).
violet heron
guess im stuck ... for some unknown reason i can not train anymore with applio -...
applio moment
crude bolt
i ended up reinstalling the whole thing - it just refuses to do the last step - the training
violet heron
i ended up reinstalling the whole thing - it just refuses to do the last step - ...
Why not use mainline
crude bolt
cuz i heard it was no longer maintained.
i did so in the past - it was also compared to applio a struggle to setup.
violet heron
i did so in the past - it was also compared to applio a struggle to setup.
Unzipping and opening a bat file?
crude bolt
welp can try again - maybe in time things changed.
actually i already got it installed
🤦♀️
violet heron
cuz i heard it was no longer maintained.
Applio has ZERO real extra features over mainline
Or at least to my testing
crude bolt
go-web.bat letsgooo
glacial pollen
welp can try again - maybe in time things changed.
log?
aside, yeah as tim said
there's no real reasons to use applio over mainline aside some utils maybe or gimmicks
one could argue about code optimizations or whatever but it's quite marginal
crude bolt
well lets see if this thing is magically doing it for me 🙂
glacial pollen
you'd actually gain more using my fork as it's mainline with my enhancements:
if I was to be honest with you
crude bolt
oh it cannot use a pre-train ?
glacial pollen
I did move from mainline to fumiama's base ( updated it ) but it's not pushed yet
however, what is available, still works so
Nah, it can use pretrain
you just gotta replace the G / D paths
crude bolt
...
ok where is that located (i feel another facepalm moment lol ... is it this easy XD )
crude bolt
... my brain is fluid / cooked atm - ok tnx
glacial pollen
np
crude bolt
🤦♀️ 🤦♀️
glacial pollen
ps, in case you intended to use my fork, clone the repo - ignore the release
thin stump
Wait this is a bit unrelated but I saw somewhere tht colab stopped giving access...
I recommend kaggle unless you have a GPU
I wrote up a quick one here but just copied over a few info from the mainline colab guide https://rentry.co/RVC-Mainline-Kaggle
plain fulcrum
I recommend kaggle unless you have a GPU
It works with Cpu right
rare gobletBOT
Ayo? @plain fulcrum level 4 !!!
thin stump
How does tht happen😭
12 hours per day which I should have switched to long ago . I got too comfortable with rvc disconnected
plain fulcrum
Oh so it runs without the need to have a gpu right
Cus I only got integrated graphics
thin stump
yup, you just need to verify your phone number and turn on the internet on kaggle
plain fulcrum
Yeah imma check that one out tmr I appreciate it
azure marshBOT
violet heron
oh it cannot use a pre-train ?
You can replace the files
oblique gust
abcdef
rare gobletBOT
Ayo? @oblique gust level 1 !!!
oblique gust
you
sick junco
i need an AI voice conversion tool where i can drop my audio file and then it will convert it
not live audio conversion
real igloo
please help, i've been fucking around with this for the past couple of HOURS and i cant figure this out https://i.imgur.com/MqnZPPo.png
i dont know what to do
like i genuinely have 0 clue what to do and how to fix this and it's so infuriating
can anyone genuinely help i swear to GOD himself i'm about to strike down my entire setup and destroy everything in it
this might be the most annoyed i've ever been by a single program on earth
and i've had issues like this before and the fixes were so goddamn easy and for some reason i COMPLETELY forgot all of my python and rvc knowledge
the brainwiping martians wiped my fucking memory
sick junco
i need an AI voice conversion tool where i can drop my audio file and then it will convert it
not live audio conversion
willow bloom
why is it laggy on v auido cable?
willow bloom
why is it laggy on v auido cable?
no mine is messed up
sick junco
oh
mild dock
Question guys
would it make the audio better if i re enter the exported file again into the output?
or is that going to distort it
sick junco
,i need an AI voice conversion tool where i can drop my audio file and then it will convert it
not live audio conversion
does anyone have one
void salmon
Hello here ! Sorry to ask. Is there a GUI that works ? Ilaria doesn't anymore for me, and Easy GUI is down for a while...
violet heron
i need an AI voice conversion tool where i can drop my audio file and then it wi...
RVC
subtle cedar
someone help me how to upload models on 'voice-models' and how to ask permission on moderators?
violet heron
someone help me how to upload models on 'voice-models' and how to ask permission...
#1159514067187277865 /submit
subtle cedar
<#1159514067187277865> /submit
It's an alternative for 'voice-models' to upload models there or anyone there can help to upload models on 'voice-models'?
violet heron
It's an alternative for 'voice-models' to upload models there or anyone there ca...
#1159514067187277865 /submit
You have to submit one of your models and get approved to post in there
subtle cedar
You have to submit one of your models and get approved to post in there
Oh okay thanks
subtle cedar
You have to submit one of your models and get approved to post in there
Incase if you're in weights's server too and if you used weights can you help me to upload models on 'model-sharing' channel from weights's server too?
violet heron
Incase if you're in weights's server too and if you used weights can you help me...
I’m not in weights server
subtle cedar
I’m not in weights server
Okay thank you
subtle cedar
You have to submit one of your models and get approved to post in there
But unfortunately I can't remember the epochs of my models because I trained it from weights not from Google Colabs since weights doesn't show the epochs for trained models there so probably I can't upload/share my models there unless I type the epochs of the models ✌️
pulsar anchor
Nah abandon the attempt now
I'll never create a very good ai voice model for fnf
Fuck my life
sick junco
someone please answer this, holy shit. i need an AI voice conversion tool where i can drop my audio file and then it will convert it
not live audio conversion
thin stump
someone please answer this, holy shit. i need an AI voice conversion tool where ...
It's the best one 
brittle wing
Applio can be good too
or RVC Mainline but both are very good, samething about Ilaria RVC
frank tinsel
Guys, I'm looking for a AI that can make a cover of Look At Me my XXXTentacion with the voice and musical style of George Straight. What discord channel or server is capable of doing this?
median monolith
when exporting an audio from a flp (fl studio project), are these the best options if i want the highest quality and fidelity possible of the audio?
the audio/sample im exporting has a samplerate of 44100Hz and has a bit depth of 16
i know there is problems when exporting an audio/sample with certain specifications if they have a certain samplerate and bitdepth, i just dont know what are the best ones for this specific audio/sample
knotty moth
when exporting an audio from a flp (fl studio project), are these the best optio...
ideally 32-bit float for mixing/mastering materials, but as a final product to publish, flac 16-bit
urban cradle
Is there a way to make another model for another artist that’s not on here if that made sense
violet heron
Is there a way to make another model for another artist that’s not on here if th...
That’s called training a model
-docs
urban cradle
That’s called training a model
Yeah it’s just complicated
violet heron
Yeah it’s just complicated
Then learn so it’s not
We have guides that show you step by step
Though if you want to be lazy we have a #1159289738314919936 channel but no one really takes requests
urban cradle
To much
violet heron
Though if you want to be lazy we have a <#1159289738314919936> channel but no on...
Or alternatively you can check if the models exist on weights gg
plain fulcrum
😭
plain fulcrum
Guys I had a question, is there like a ratio for the length of the dataset and epoch number?
Like yesterday I trained something as a test with 1 min of data and 50 epochs, so if today I wanna increase the data to 10 mins, how much should I increase my epochs by
pliant stream
Like yesterday I trained something as a test with 1 min of data and 50 epochs, s...
Sorry i misunderstood ignore my previous messages.
I would train at about 150-250 epochs for 10 minutes data if you use the default Pretrain
brittle wing
MY RVC on google collab is not working
carmine hearth
-colab
plain fulcrum
How long does 1 epoch take for u guys on average?
For me it took around 1 hour and a half for 50 epochs
plain fulcrum
Sorry i misunderstood ignore my previous messages.
I would train at about 150-25...
Yeah imma try 150 epochs today when I get my 10 min dataset
simple crater
So how do people handle harmony? the AI has trouble picking a pitch and jumping around in background vocal tracks
any fix or remediation?
plain fulcrum
Do the epochs speed up during the training?
glad zealot
Do the epochs speed up during the training?
nope
its mostly around the same time as the first epoch
sand glade
помогите установить
rare gobletBOT
Ayo? @sand glade level 1 !!!
low shard
I wrote up a quick one here but just copied over a few info from the mainline co...
Bro fr made a guide
low shard
Sorry i misunderstood ignore my previous messages.
I would train at about 150-25...
Also putting like 500 epochs and using tensor board is good
low shard
MY RVC on google collab is not working
!howtoask
steady bolt
Hello so, question. my RES keeps increasing but only for RVC models. I'm using Voice Changer and I just got into it so I'm messing with the models that are already in the program. I've tried the Beatrice model and it works fine but for RVC models the RES keeps rapidly increasing.
how do I make it stop constantly increasing
I've tried lowering the chunks and extras and I'm currently at 4096 for extra which is the lowest and 4 chunks
wispy lodge
You should increase chunks, not lower them
steady bolt
oh thanks!
red cliff
Hey if i don't use pitch guidance when training a rap model, what decides pitch? It sounds good, but it's interesting because it does seem to be changing pitch (its not a single note), but it's not following the input audio
violet heron
Try adjusting the pitch
Yeah but it will sound less gay if you adjust the pitch
Be straight then
proper shale
😭
median monolith
ideally 32-bit float for mixing/mastering materials, but as a final product to p...
I see. so i can export this 44100hz samplerate and 16 bitdepth sample with the settings that i showed and it will export perfectly as i hear it in the flp? (im afraid to export it with wrong configurations and mess with the frequency (make it higher pitched) or make it not the highest quality and fidelity possible for it), do i need to lower a certain option (like the bit float one) to export it with no artifacts or problems at all?
scarlet wedge
hi
what is the version rvc ilaria is using?
huggingface
the version number
or the gradio version number
rare gobletBOT
Ayo? @scarlet wedge level 4 !!!
delicate wind
hey im getting this error in google colab
[Errno 2] No such file or directory: '/content/voice-changer/server'
/content
ModuleNotFoundError Traceback (most recent call last)
<ipython-input-2-cc642795cb8e> in <cell line: 24>()
22 get_ipython().run_line_magic('cd', '/content/voice-changer/server')
23
---> 24 from pyngrok import conf, ngrok
25 MyConfig = conf.PyngrokConfig()
26 MyConfig.auth_token = Token
ModuleNotFoundError: No module named 'pyngrok'
NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.
To view examples of installing some common dependencies, click the
"Open Examples" button below.
analog obsidian
hey im getting this error in google colab
[Errno 2] No such file or directory:...
which colab are u using?
unreal lotus
Is there a new google colab link? If so, could anyone send me it? Thanks
analog obsidian
Is there a new google colab link? If so, could anyone send me it? Thanks
for realtime voice changer or rvc?
unreal lotus
Rvc
analog obsidian
Rvc
nope, every colab is updated now
when a colab gets updated/moved to other link will be announced on this channel: #📰│dev-updates
analog obsidian
I had an rvc v2 lol
-colab
hollow fox
guys how can ı run .sh files ? ı tried bash but it didn't worked
brittle wing
Is there a tutorial for making a model from start too finish?
low shard
Is there a tutorial for making a model from start too finish?
yep, check https://docs.aihub.wtf/essentials/how-to-make-voice-models/ for rvc models
brittle wing
applio colab isn't making a public url
low shard
applio colab isn't making a public url
use the 'share_tunnel' option and re run the start applio cell
brittle wing
use the 'share_tunnel' option and re run the start applio cell
okay i couldn't figure out what to do with that at first and thought it also wasn't working but now i know how to use it
low shard
okay i couldn't figure out what to do with that at first and thought it also was...
Alr seems like it works fine, if u get other errors u can ask here
fierce hinge
violet heron
Where did you get this file from?
fierce hinge
violet heron
That is a model file
fierce hinge
yes
violet heron
You need to use a program like RVC or w-Okada to use it
fierce hinge
any tut?
violet heron
-realtime
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
violet heron
First one
fierce hinge
/guides realtime
violet heron
Should have settings, downloads, tips, and basically everything else
Ignore the rickroll
violet heron
there is no yt tut?
YouTube tutorials often are outdated and spread misinformation
Like I’ve seen YouTubers say you need to download python/pytorch, which you don’t need to since it’s bundled with it
fierce hinge
um
violet heron
Which is for ai covers, not realtime
2nd link is for real-time RVC not w-okada
Which you can use instead but w-okada has better ui and is easier to use
fierce hinge
i just want a realistic girl voice in realtime for fivem 
violet heron
<https://rentry.co/VoiceChangerGuide>
Use this one
It’s really just downloading, unzipping, running a file and selecting the model and settings
Which you already have the model
fierce hinge
okay
idk where to click on the guide xd
some github link some nvdia cuda
idk xd
fierce hinge
violet heron
Here’s the direct link
rare gobletBOT
Ayo? @fierce hinge level 2 !!!
violet heron
Set input to your microphone
Set output to the audio cable (see guide)
Set monitor to your speakers/headphones
Select your GPU as the GPU
fierce hinge
audio cable?
violet heron
audio cable?
In order to use the voice changer in other apps you need to get a audio cable
vac lite is the best free one that most people use
violet heron
i used it for voicemetter
that one would also work
fierce hinge
kk
violet heron
People have reported it being kinda buggy tho, so if your experiencing issues with audio in other apps with it use the one I linked
fierce hinge
yes
then when i do that
how can i heard myself
in real time?
to see the result
violet heron
in real time?
Select a model then click start
fierce hinge
i have to change some setting to get the voice realistic?
violet heron
Select a model then click start
To add a model click edit, then click upload, then upload the .pth and .index that were in the models zip file
violet heron
i have to change some setting to get the voice realistic?
Tune
Also make sure to use RMVPE
fierce hinge
whats that?
violet heron
whats that?
A setting under f0 det:
@fierce hinge also make sure to use the chunk and extra the guide says to for your GPU
fierce hinge
fierce hinge
<@1118975004277743718> also make sure to use the chunk and extra the guide says ...
what i have to change?
so i put rmvpe
fierce hinge
wtf
my model is weird
not even looking like the preview
eww
OHGHH
its sound cool
how to stop heard myself xd
violet heron
how to stop heard myself xd
Set monitor to none
violet heron
I personally keep it so I know the changer isn’t acting up
violet heron
but it sound weird on discord
Does it sound weird on the voice changer itself?
fierce hinge
a bit
fierce hinge
yes
violet heron
Might be the audio cable
violet heron
what chunk
112
fierce hinge
it crashh
fierce hinge
rare gobletBOT
Ayo? @fierce hinge level 3 !!!
violet heron
K
violet heron
https://software.muzychenko.net/freeware/vac470lite.zip
Try using this audio cable instead
See if that sounds better on discord
fierce hinge
Try using this audio cable instead
im using this one
violet heron
Increase chunk and it should be good
fierce hinge
increase?
violet heron
Might want to turn off index if there’s weird artifacting
fierce hinge
how much?
violet heron
its good in real time but sum sounding weird
Unmute your mic so I might be able to hear what’s wrong
Might be the tune
violet heron
i dont really know
Sounds good to me tho
fierce hinge
u think so ?.??
violet heron
Say soemthing again
fierce hinge
might be the discord mic test
violet heron
might be the discord mic test
Yeah
violet heron
so everything should be good
Should
fierce hinge
kk
fierce hinge
okay
w w w
ill test on fivem later
but something is weird
liike
its look like a accent
violet heron
its look like a accent
Accent is controlled by the index
fierce hinge
owww
violet heron
ill test on fivem later
Just a little note you will need to limit frame rate or game settings
So you can give more room for the voice changer to convert and stuff
fierce hinge
Just a little note you will need to limit frame rate or game settings
kk
and when i start to talk it make like
pp
pp
like idk
maybe only on the client
fierce hinge
and sometime everything work
then a weird voice come xd
idk wtd
and what is passthru
violet heron
and what is passthru
Disables the voice changer and uses your real voice
fierce hinge
kk
brittle wing
-colab
urban wasp
Hey guys, ı have problem with applio goggle collab..
I can't downland the file
Or the audio doesn't play when ı press the button
violet heron
Or the audio doesn't play when ı press the button
Did you convert it
urban wasp
Yep
Many times
Im not using the collab right now, but ı got that problem multiple times
violet heron
applio moment
urban wasp
And ı don't know why, when it start to downland, it took so long even it's just 5mb
main tusk
will my pc work well with the ai thing.
i have a radeon 6900xt 32 gb ram 12th gen i7-2700k
urban wasp
I made these from mobile
lime otter
brittle wing
So, whenever I save an audio I generate with Applio, it downloads but the whole page resets. I'll have to replicate the error and look at the cli window but that's basically what's happening right now
it never did this before for some reason
Ahem...no one is ready to hear this but believe me : voice conversion and deepfakes is nothing new...it existed a few years ago, it's just now that's accessible to the public for personal use
digital stump
someone has a google colab for real time voice changer
violet heron
someone has a google colab for real time voice changer
-rt
azure marshBOT
-rt
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
digital stump
thx
brittle wing
someone has a google colab for real time voice changer
unreal basalt
WHAT THE FU-
unreal basalt
WHAT THE FU-
First Kits, then Weights!?!?
violet heron
First Kits, then Weights!?!?
Just use RVC
thin stump
Just use RVC
they get way too comfortable with sticking on one platform and don't want to learn anything new 
. Can't force them since that anger issue guy also ignored the pings
violet heron
they get way too comfortable with sticking on one platform and don't want to lea...
For some reason that bro doesn’t want to use RVC when it’s basically the exact same thing with different ui
People don’t want to experiment since they want their ai e girl as fast as possible
thin stump
I can fix you nah, you're cooked
void stratus
its not letting me launch it anymore, when i click it nothing happens
violet heron
its not letting me launch it anymore, when i click it nothing happens
Is everything unzipped?
void stratus
Is everything unzipped?
yes
void stratus
nevermind i found a differnet file with the exact name and that ones working
do yk how to upload a voice effect to the thing?
rare gobletBOT
Ayo? @void stratus level 2 !!!
knotty moth
its not letting me launch it anymore, when i click it nothing happens
try this if it closes instantly #🔍│help-w-okada message
late raptor
hey guys
so, i cant actually upload files to gradio
it just wont show up at the dropdown menu
i'm using Ilaria-Mainline
can someone help me?
i don't really know what to do
violet heron
i'm using Ilaria-Mainline
Click refresh
violet heron
do yk how to upload a voice effect to the thing?
“Edit”
late raptor
i've did it a lot
late raptor
Click refresh
still won't show up
violet heron
still won't show up
Wait are you using Ilaria mainline locally?
violet heron
not really
Wdym not really
Are you using Ilaria RVC hf space or did you download it and run a file
Or are you using a colab
late raptor
because i dont know how to use it locally
violet heron
because i dont know how to use it locally
Im asking what platform your using it on
late raptor
colab
violet heron
Are you trying to make a cover or model
late raptor
cover
violet heron
cover
I would recommend to use the current HuggingFace space for Ilaria RVC instead
It’s way faster and has easier ui
-spaces
late raptor
thank you sm
i actually didnt know wich one to use
thanks
ive been having a hard time with colab lately
damn it IS better
actually easier to use
void stratus
so i have the app and stuff, but when i speak in the app or in a external app like discord i cant hear myself speaking, it doesnt detect a audio
violet heron
so i have the app and stuff, but when i speak in the app or in a external app li...
-realtime you need to get a audio cable
azure marshBOT
-realtime you need to get a audio cable
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
void stratus
i already got one
after i open up start_http it leads me to BOTH the website and the app, but before it only lead me to the website and it was working fine
after leading me to both the website and the app if i close the app the website automatically closes
void stratus
-realtime you need to get a audio cable
okay so its working in thea pp i can hear myself
violet heron
okay so its working in thea pp i can hear myself
Is your output on w-Okada set to the cable?
void stratus
nope ij ust made it so its default
it was cable before for the output
but it wasnt working
violet heron
Ilaria RVC Zero is recommended now for online inference
violet heron
Output should be the cable and monitor should be your headphones
violet heron
Ilaria RVC Zero is recommended now for online inference
-spaces @faint bloom
azure marshBOT
-spaces <@821335987329564682>
Suggestions for @faint bloom
void stratus
Output should be the cable and monitor should be your headphones
aha thank you that works
knotty moth
easygui has been discontinued, Ilaria RVC is the best bet for inference
https://docs.google.com/document/d/1YbXcLFPaGjhOdG5NFkK3QrucCEpHZBwFUxkeMO8aB18/edit
spice thorn
How to use tts?
plain fulcrum
Guys how do you deal with sfx in the dataset? Im making a dataset for plankton but anytime he speaks there's some random cartoon sfx that stick even after I isolate the vocals, is there any way to remove em?
stuff like this
i could hear these sfx in the vocals when i was testing the model
pastel oak
stuff like this
Spectral denoise with rx izotope maybe
else try mvsep.com and use the bandit one i forgot where it was exactly
would try mvsep first
plain fulcrum
Thx so much I'm ganna try it out
pastel oak
Worst case cut out all the words where the noise happens
pastel oak
Thx so much I'm ganna try it out
Oh well i guess you can also try BS-Roformer and use 2024.04 on mvsep
usually does the trick, i forgot about it for a sec
spice thorn
help with tts pls it generating tts but not convert to ai model voice
rare gobletBOT
Ayo? @mint mulch level 1 !!!
proven hill
help with tts pls it generating tts but not convert to ai model voice
use it on ilaria rvc
ember shore
hey guys im very confused, what is titan? is it the new rmvpe? and how do i use it?
proven hill
hey guys im very confused, what is titan? is it the new rmvpe? and how do i use ...
its a pretrain
ember shore
What does that mean?
proven hill
its a thing for training models, dont bother
ember shore
Oh okay thankyou
rare gobletBOT
Ayo? @ember shore level 1 !!!
ember shore
One more question tho
Does it matter what I use titan on? I mean is it better or mango crepe or rmvpe, that sort of thing
lunar ice
the easy Gui website is not working?
proven hill
Does it matter what I use titan on? I mean is it better or mango crepe or rmvpe,...
its not a rmvpe or mangio crepe thing (its called pitch extraction method what youre saying), completely different thing
proven hill
the easy Gui website is not working?
can you send the link?
lunar ice
can you send the link?
proven hill
https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sh...
thats super old
proven hill
lunar ice
Is there a tutorial to learn how to use this?
rare gobletBOT
Ayo? @lunar ice level 1 !!!
proven hill
Is there a tutorial to learn how to use this?
hun the link i sent you is the tutorial
lunar ice
oh sorry
proven hill
no prob bbg
spiral whale
Hi! I'm trying to make a voice model for Cynthia Erivo because I need her voice in a couple phrases and a bit of a song for a small performance. I am at the phase of isolating her vocals in a bunch of songs but I'm very confused